Augmented Ops - Episodes Tagged with “Ai”

Hacking the Defense Bureaucracy: Software, Speed, and the Industrial Base

Tulip — Thu, 14 May 2026 00:15:00 -0400

Geopolitical pressure is reshaping how the US buys and builds for defense. Nick Sinai of Insight Partners breaks down the shift from cost-plus to commercial procurement, the rise of venture-backed defense tech, and how to move fast without losing safety.

In this episode of Augmented Ops, host Erik Mirandette, Tulip's Chief Business Officer, is joined by Nick Sinai, Managing Director at Insight Partners and co-author of Hack Your Bureaucracy. Before Insight, Nick spent nearly six years inside the Obama administration as US Deputy CTO, where he led the Open Data Initiative and helped stand up the Presidential Innovation Fellows program.

Nick breaks down what's actually changing in defense procurement under the second Trump administration, the rise of venture-backed defense tech now drawing tens of billions of dollars a year, and the shift from cost-plus to commercial-products buying. He explains why the traditional 15-year acquisition cycle no longer matches the pace of technology, and how companies like Anduril, Shield AI, and Palantir are reshaping what it means to be a defense prime.

The conversation also explores the tradeoff every modern supplier into defense has to navigate: how to move faster on cost and speed without junking the safety and compliance requirements that exist for good reason. Nick offers a grounded view from someone who's lived inside both the bureaucracy and the venture world, including what it actually takes for reform to stick across administrations.

The headlines are full of speeches about speed. Nick lays out what's genuinely different this time, what's likely to regress to the mean, and where operations leaders supplying into the defense base should be paying attention.

Watch the full episode on YouTube: https://youtu.be/Qu7bCfNpraA

Augmented Ops is a podcast for industrial leaders, citizen developers, shop floor operators, and anyone who cares about what the future of frontline operations will look like across industries. This show is presented by Tulip, the Frontline Operations Platform. You can find more from us at Tulip.co/podcast or by following the show on LinkedIn. Special Guest: Nick Sinai.

AI for Operations: From Everyday Tools to Agentic Systems

Tulip — Thu, 30 Apr 2026 00:00:00 -0400

Manufacturing is entering a new phase of AI adoption, one where intelligent systems don’t just generate insights but take action in context. In this episode of Augmented Ops, host Mason Glidden, Tulip’s Chief Product Officer, is joined by Olga Stroilova, Group Product Lead, and Pete Hartnett, Group Product Manager, to discuss how agentic AI is redefining what’s possible on the factory floor.

Together, they unpack the evolution from predictive and generative AI to agentic systems capable of autonomous, goal-driven behavior while keeping people firmly in the loop. They examine why many pilots stall before production, how governance and culture shape adoption, and why “human oversight by design” is becoming the new standard for responsible AI in manufacturing.

Drawing from Tulip’s own roadmap and customer experiences, the team highlights how features like AI Composer, Tulip Agents, and context-aware workflows are helping users close the insight-to-action gap, scale AI safely, and unlock new forms of operational leverage.

Rather than imagining a future without people, the episode points to a more realistic vision of AI in manufacturing: one where systems evolve, but human judgment remains the foundation of progress.

Augmented Ops is a podcast for industrial leaders, citizen developers, shop floor operators, and anyone else that cares about what the future of frontline operations will look like across industries. This show is presented by Tulip, the Frontline Operations Platform. You can find more from us at Tulip.co/podcast or by following the show on LinkedIn.
Special Guests: Olga Stroilova and Pete Hartnett.

The Physics Layer: Why AI Needs Real-World Engineering to Unlock Trillion-Dollar Industrial Value

Tulip — Thu, 16 Apr 2026 00:15:00 -0400

What do a floating barge the size of four aircraft carriers, a Shell refinery, and the future of energy resilience have in common? They all depend on knowing, with precision, how much life is left in the steel.

In this episode of Augmented Ops, host Natan Linder sits down with Thomas Leurent, CEO and co-founder of Akselos, to unpack the often-overlooked world of structural performance management and why it might be the most important form of physical AI you've never heard of.

Thomas shares how Akselos helped Shell unlock over half a billion dollars in value on a single FPSO by using physics-based digital twins to skip a costly dry dock. He explains the technology behind it: a proprietary approach that runs structural simulations 100,000x faster than traditional finite element analysis by blending machine learning with physics, built over 20+ years since the technology was pulled out of MIT.

The conversation goes deep on what physical AI actually means in industrial settings, why hallucination is simply not an option in high-stakes environments, the role of humans in process industries (especially in emergency scenarios like what's unfolding in the GCC), and how data sharing — or the lack of it — is holding back offshore wind and the broader energy transition.

Thomas also shares a bold prediction: just as algorithmic efficiency transformed mechanical simulation, it will do the same to AI, making large language models far cheaper to run, potentially leading to an overcapacity of computing infrastructure in the years ahead.

If you think "structural performance management" sounds dry, wait until you hear what a $500M dry-dock skip, 52,000 workers with zero casualties, and the future of energy supply chains have to say about it.

Watch the full episode on YouTube: https://youtu.be/cmhRmhyrU-Y

Modernizing the Industrial Base: Readiness, Resilience, and the Road Ahead

Tulip — Thu, 02 Apr 2026 00:15:00 -0400

Modernizing the U.S. industrial base is no longer a long-term goal. Between geopolitical competition, workforce constraints, and rising cybersecurity demands, manufacturers are under pressure to rethink how production systems are built and operated.

Clark Dressen, CTO of MxD, joins the show to explain how this transformation is taking shape across the defense industrial base and broader manufacturing ecosystem. As a public-private partnership funded in part by the Department of Defense, MxD works to connect emerging technologies with real-world production environments.

The conversation focuses on what modernization actually requires. Not digital transformation for its own sake, but applying technology to solve specific operational problems around quality, productivity, and consistency. Clark shares how tools like sensors, digital twins, and AI are being introduced into legacy environments to reduce reliance on tribal knowledge and create more repeatable processes.

The episode also explores the structure of the industrial base, where small and mid-sized suppliers make up the majority of the defense supply chain but often lack the resources to meet growing cybersecurity and compliance requirements. As workforce transitions accelerate, the focus shifts toward capturing expertise, improving how work is executed, and building more resilient production systems.

Watch the full episode on YouTube: https://youtu.be/T2bkZvyK5kU

Regional Ecosystems of Manufacturing: The Foundation of Industrial Strength

Tulip — Thu, 19 Mar 2026 00:15:00 -0400

What actually makes a region strong in manufacturing?

In this episode, Gillian Catrambone sits down with Beatriz Gutierrez, CEO of CONNSTEP Inc., Connecticut’s Manufacturing Extension Partnership (MEP), to explore how regional ecosystems, through MEPs, workforce programs, and coordinated resources, create the foundation for industrial strength.

Beatriz breaks down how manufacturers are navigating labor constraints, adopting automation incrementally, and rethinking supply chains in a more volatile environment. The conversation also highlights what separates effective regions, including strong talent pipelines, connected institutions, and easier access to capital, training, and support.

She closes with practical guidance for operations leaders. Focus on critical processes, plan for the long term, and approach transformation step by step rather than waiting for perfect conditions.

Watch the full episode on YouTube: https://youtu.be/ZJO0bbYSGII

The State of Reshoring: Supply Chains, Strategy, and the Future of US Manufacturing

Tulip — Thu, 05 Mar 2026 00:15:00 -0500

Global supply chains are being rewired in real time. From tariffs and geopolitics to labor constraints and energy infrastructure, manufacturers are navigating a level of volatility few have experienced before.

Rosemary Coates, Founder and Executive Director of the Reshoring Institute, joins the show to unpack what’s actually happening beneath the headlines. Drawing on recent executive interviews and location studies, she explains why many companies are pausing major decisions, how “China plus one” strategies are evolving, and what reshoring really requires beyond political rhetoric.

For operations leaders, the conversation moves from macro forces to practical considerations: evaluating total landed cost beyond labor, balancing capital intensity with workforce availability, selecting locations with infrastructure in mind, and building resilience through diversified manufacturing footprints. While the path forward is complex, Rosemary outlines why advanced, higher-skilled manufacturing still presents meaningful opportunity for U.S. growth.

Watch the full epsiode on YouTube

For more on this topic, Rosemary hosts The Frictionless Supply Chain podcast, covering supply chain strategy and global production shifts.

From the Classroom to the Shop Floor: Building the Future Industrial Workforce

Tulip — Wed, 18 Feb 2026 00:15:00 -0500

Manufacturing is undergoing a generational shift. As experienced workers retire and automation accelerates, the industry must solve both a workforce shortage and a skills gap — and it must do so simultaneously.

Jacob “MFG Kid” Sanchez is a well-known manufacturing influencer and content creator, and a vocal advocate for bringing new talent into the industry. With hands-on shop floor experience and a growing platform dedicated to promoting automation and modern manufacturing careers, he works to make the industry more visible, accessible, and appealing to the next generation.

Check out Jacob’s newly launched Axis community — a brand-neutral space for automation, robotics, and manufacturing professionals to connect, learn, and collaborate.

In this conversation, Jacob and Natan explore how manufacturers can generate genuine interest in industrial careers, rethink how technical skills are taught and developed, and draw lessons from apprenticeship models in countries that consistently produce highly skilled manufacturing talent.

Watch the full episode on YouTube

The Need for Speed in Life Sciences

Tulip — Thu, 05 Feb 2026 00:15:00 -0500

The life sciences industry has long justified slow digital adoption through regulation. But as technology adoption accelerates faster than guidance, that logic is breaking down.

In this episode, Michelle Vuolo and Gilad Langer discuss why speed has become a defining challenge for pharma and medical device manufacturers. Drawing on experience from ISPE, quality leadership, and decades in regulated operations, they explore validation 4.0, cultural resistance to risk-based thinking, and how AI is reshaping quality and compliance work.

The conversation examines what it really takes for life sciences organizations to move faster without losing control — and why waiting for perfect regulatory clarity is no longer a viable strategy

Watch the full episode on YouTube: https://youtu.be/SPJz8_cFYM4

The Human Infrastructure of Manufacturing with Stacey Weismiller of AMFI

Tulip — Thu, 22 Jan 2026 00:15:00 -0500

Manufacturing is often discussed in terms of technology, productivity, and investment — but rarely in terms of people as infrastructure. In this episode of Augmented Ops, Stacey Weismiller, Founder of the American Manufacturing Futures Institute, joins Natan Linder to reframe the conversation.

Stacey draws on her background spanning manufacturing, economic development, and global policy to explore why people, access, and community must sit at the center of industrial renewal. Together, they discuss workforce participation, civic manufacturing, equitable growth, and how AI can augment human work without eroding trust or dignity.

The conversation spans everything from factory jobs and childcare to resilience, reindustrialization, and why manufacturing needs a new narrative — one that values stewardship as much as efficiency.

Watch the full episode on YouTube.

The Real Problem AI Needs to Solve in Manufacturing

Tulip — Thu, 15 Jan 2026 09:00:00 -0500

Read about Tulip’s $120M Series D 👉 http://tulip.co/press/tulip-secures-120m-series-d/

In this conversation, Chris Luecke, host of the Manufacturing Happy Hour podcast, joins Natan Linder to discuss AI in manufacturing, and how Mitsubishi Electric’s lead investment in Tulip’s $120M Series D helps to accelerate our mission to scale our composable platform, support an open ecosystem for frontline operations, and supports AI-enabled and human-driven innovation.
Key themes from this conversation include:
• Why "software-defined manufacturing" is essential for modern supply chains.
• The rise of the AI process engineer, and real-world implications of AI adoption among frontline process engineers.
• The importance of building a transparent, human-first culture in frontline operations.

AI at the Crossroads of Regulation and Innovation

Tulip — Thu, 08 Jan 2026 12:00:00 -0500

AI is rapidly reshaping life sciences manufacturing—but as intelligent systems move into regulated environments, questions around validation, governance, and trust become unavoidable.

In this episode of Augmented Ops, host Michelle Vuolo, Head of Quality at Tulip, is joined by Bryan Ennis, Chief Quality Officer and Founder of Sware, and Martin Heitmann, of the Triality Group. Together, they explore what it really takes to deploy AI responsibly in pharma, biotech, and medtech operations.

The conversation examines why many AI initiatives stall at the pilot stage, how validation practices must evolve for probabilistic systems, and where organizations are already seeing real value—from predictive maintenance to quality signal detection and validation automation. They also discuss emerging regulatory guidance, including Annex 22, and why regulators are not anti-AI—but deeply skeptical of black-box systems.

Throughout the discussion, a consistent theme emerges: successful AI adoption is less about the technology itself and more about process design, data quality, human oversight, and building evidence that systems are safe, transparent, and fit for purpose.

This episode offers a grounded, experience-driven perspective on how life sciences organizations can move from experimentation to scale—without compromising patient safety or compliance.

Watch the full episode on YouTube: https://youtu.be/7keK_4zDaTg

Giving Robots "Common Sense": Inside RightHand Robotics with Yaro Tenzer

Tulip — Thu, 18 Dec 2025 00:15:00 -0500

Robotics has promised to transform manufacturing and logistics for decades — but turning intelligent machines into reliable, everyday operators remains hard. In this episode of Augmented Ops, Natan Linder sits down with Yaro Tenzer, co-founder and CEO of RightHand Robotics, to talk about what it actually takes to deploy AI-powered robotics in real operational environments.

Yaro shares lessons from building robotic systems that operate in the messiness of the real world — where data is imperfect, edge cases are constant, and reliability matters more than demos. Together, they discuss why so many robotics pilots struggle to reach production, how machine learning improves through real-world feedback, and what operations leaders should understand before investing in automation.

The conversation explores the intersection of robotics, AI, and operations — focusing on practical constraints, system design, and the human decisions that determine whether advanced technology delivers value or stalls on the shop floor.

Watch the full episode on YouTube: https://youtu.be/a06GA7TvI8Y

The Future Process Engineer with Chris Luecke of Manufacturing Happy Hour

Tulip — Thu, 04 Dec 2025 00:15:00 -0500

What does the future of process engineering look like in an era shaped by AI, automation, and rapid operational change? In this episode, Chris Luecke joins Natan Linder to explore how the role is evolving, what still defines great engineering, and why human judgment remains essential on the modern shop floor.

Chris is the host of Manufacturing Happy Hour and one of the most connected voices in the industry. Before stepping behind the microphone, he spent years as a process engineer at Anheuser-Busch and later worked across sectors with Rockwell Automation—giving him a rare vantage point on how factories actually run and how engineering teams solve problems.

Natan and Chris discuss the shift from reactive troubleshooting to systems thinking, how culture shapes the pace and quality of improvement, and why the most effective way to introduce AI is to aim it at the tasks teams collectively find painful. They also examine the idea of “Shenzhen Speed,” how faster design and production cycles influence global competitiveness, and what manufacturers elsewhere can learn from regions that move quickly.

This conversation offers an on-the-ground view of how engineering work is changing and what the next generation of process engineers will need to thrive.

Watch the full episode on YouTube

AI, Industry, and the Human Story with MIT’s David Mindell

Tulip — Thu, 13 Nov 2025 00:15:00 -0500

AI is often described as a revolution, but every technological leap has deep roots in the human story. In this episode of Augmented Ops, MIT Professor and author David Mindell joins Tulip CEO Natan Linder to discuss how history can help us navigate the rise of intelligent systems.

Mindell, a historian, engineer, and entrepreneur, shares insights from his latest book, The New Lunar Society, which traces the origins of the Industrial Revolution and the people who built it. He draws connections between the 18th-century innovators who powered the first era of mechanization and today’s engineers shaping AI. Every tool, he argues, embeds human skill, judgment, and culture; from the earliest steam engines to modern autonomous systems.

Their conversation examines the enduring questions that define manufacturing and technology: How can new tools expand opportunity instead of narrowing it? What does responsible innovation look like in an age of automation? And how can societies balance ambition, governance, and trust while embracing change?

Through stories of invention, work, and rediscovery, Mindell reminds us that progress has always been a human endeavor. Technology evolves, but the drive to create, understand, and improve remains constant.

Watch the full episode on YouTube

AI for Operations: From Everyday Tools to Agentic Systems

Tulip — Thu, 30 Oct 2025 00:15:00 -0400

Rather than imagining a future without people, the episode points to a more realistic vision of AI in manufacturing: one where systems evolve, but human judgment remains the foundation of progress.

Operations Calling 2025 Recap: From AI Hype Into Real World Results

Tulip — Thu, 16 Oct 2025 00:15:00 -0400

Kicking off Season 6, Natan and Tulip CMO Madilynn Castillo reflect on Operations Calling 2025—recorded just after nearly 800 manufacturing leaders, engineers, and frontline pros converged at Tulip HQ for Tulip’s biggest event to date. More than a showcase of technology, this two-day experience blended strategy, execution, and genuine community. Attendees dove into headline keynotes, fireside chats, interactive workshops, and panels, led by senior voices and industry experts driving the new era of manufacturing.

The episode captures how this convergence marked a real inflection point: AI moving from hype to hands-on tools like Tulip Agents, composable systems scaling across teams, and the shift from digital transformation to continuous transformation on the shop floor. Through live demos, open learning, and collaborative problem-solving, participants saw—and built—the next wave of operations-led innovation.

Packed with post-event momentum, Natan and Madi share stories and lessons that reveal how practical AI, human-centered design, and community are reshaping manufacturing’s future.

Check out all the Operations Calling Sessions!
Natan's Keynote
The Next Shift: AI-Driven Transformation of the Connected Factory
Tulip Roadmap Session Special Guest: Madilynn Castillo.

Reindustrializing in 2025 — AI, Scale, and the Future of U.S. Manufacturing with MIT’s Liz Reynolds

Tulip — Thu, 14 Aug 2025 00:15:00 -0400

In this bonus episode, our guest is Liz Reynolds, manufacturing and workforce expert at MIT and strategic advisor to Tulip..

Fresh from Detroit's Reindustrialize conference, Liz and Natan share key insights on America's urgent push to bring manufacturing back home. They explore the "Spring of momentum" in reindustrialization efforts, from AI moving beyond hype to real implementation on the shop floor, and break down the massive scale challenges facing US manufacturers across critical sectors.

Drawing from major industry conferences including Reindustrialize, the Hill and Valley Forum, Industry Studies Association, and MIT's Initiative for New Manufacturing, she explains strategic workforce development approaches to address the 400,000 manufacturing worker shortage and the Department of Defense's $1 trillion budget impact on industrial capacity. Reynolds sheds light on how this Spring's discussions and strategic planning around technology adoption and workforce training are beginning to take concrete shape as the real work accelerates into Fall.

AI, Agility, and Enterprise Architecture – Reflections on Season 5 of Augmented Ops

Tulip — Wed, 18 Jun 2025 06:00:00 -0400

This week marks the final episode of Season 5 of Augmented Ops! Natan and Erik look back and share their biggest takeaways from conversations with CEOs, frontline engineers, and operations leaders.

They unpack marketing hype vs. the current state of AI on the shop floor, the rise of citizen developers, and why “digital transformation” needs a serious rebrand.

They also dive deep into the demise of traditional MES, explore the shift toward composable, platform-driven architectures, and offer predictions on how adaptability will define manufacturing success in the year of the “composable operations.”

Hannover Messe Unfiltered: The Real State of AI in Manufacturing

Tulip — Wed, 09 Apr 2025 08:00:00 -0400

This week, Natan Linder, Co-Founder and CEO of Tulip sits down with Madilynn Castillo, Tulip’s CMO for a recap of their experiences at this year’s Hannover Messe—the world’s largest industrial trade fair.

They explore the overwhelming AI hype at the fair, the confusion it creates for manufacturers, and the persistent gap between marketing and real operational impact when it comes to these shiny new tools. Plus, an overview of the latest developments from Tulip that debuted at the show.

How AI Agents are Closing the Loop with Composabl’s Kence Anderson

Tulip — Wed, 18 Dec 2024 04:30:00 -0500

This week’s guest is Kence Anderson, CEO of Composabl.

Kence and Natan explore the role of AI agents in industrial processes, how manufacturers are using them to solve complex problems like scheduling or machine control, and the challenges of building startups in the manufacturing space. Kence also lays out how his concept of machine teaching differs from traditional machine learning techniques, and why he believes in taking a composable approach to building solutions for problems on the shop floor.

A New Approach to Pharma Manufacturing with Roche’s Daniele Iacovelli

Tulip — Wed, 09 Oct 2024 00:30:00 -0400

This week’s guest is Daniele Iacovelli, SVP, Global Head of Digital, Analytics (AI) & Operational Excellence at Roche.

Although regulated industries like pharmaceutical manufacturing can be slow to digitally transform their manufacturing operations, Daniele lays out how changing trends in the industry have led Roche to architect a new digital production system, with a focus on a composable, future-proof architecture. He also lays out how rapid solution development enabled by new software platforms and a citizen developer approach is enabling people across the company to solve their own problems, while also laying the groundwork for rethinking the traditional software validation approach that dominates regulated industries. Plus, what the role of GenAI should be in a GxP environment, and more.

The Past, Present, and Future of Industrial Software with Rick Bullotta

Tulip — Wed, 25 Sep 2024 00:30:00 -0400

This week’s guest is Rick Bullotta, longtime veteran of the industrial software space and co-founder of Lighthammer and Thingworx.

Rick and Natan explore the history of MES solutions, how Frontline Operations Platforms are democratizing the development of industrial software, and what this means for the future architecture of the manufacturing tech stack. They also break down the hype vs. reality in industrial AI, and discuss which use cases continue to require a human-in-the-loop.

Episode 141: Reindustrializing America with Liz Reynolds

Tulip — Mon, 01 Jul 2024 04:30:00 -0400

In this bonus episode, Elisabeth Reynolds—MIT Professor of the Practice, former White House policymaker, and now Strategic Advisor to Tulip—joins Natan Linder to discuss the building momentum around reindustrialization in the United States.

Liz calls attention to the most important factors shaping the industrial landscape, and the need for a clear national strategy that can direct government coordination with manufacturers. Liz also explores the challenges in introducing software to the frontline workforce, ways manufacturers can address skill gaps, and the role of venture capital in fueling innovation.

Rendustrialize Summit
Reindustrialize Manifesto
SCSP Advanced Manufacturing Report

Augmented Ops is a podcast for industrial leaders, shop floor operators, citizen developers, and anyone else that cares about what the future of frontline operations will look like across industries. This show is presented by Tulip, the Frontline Operations Platform. You can find more from us at Tulip.co/podcast or by following the show on LinkedIn. Special Guest: Elisabeth Reynolds.

Episode 140: A Frontline Perspective on Industry 4.0 – Reflections on Season 1 of Augmented Ops

Tulip — Wed, 26 Jun 2024 07:00:00 -0400

This week marks the final episode of Season 1 of Augmented Ops! Natan Linder and Erik Mirandette sit down to discuss their takeaways from the first season—while trying not to get derailed analogizing frontline operations to the Celtics Championship win.

Natan and Erik highlight the advancements (and stumbles) in industrial AI, and the way that open, interoperable ecosystems have fundamentally changed the way manufacturing tech stacks are built. They also reflect the need to focus on the frontline worker, the power of democratizing advanced technology, and more.

Episode 139: How to Architect Your Digital Strategy with Jeff Kramer

Tulip — Wed, 12 Jun 2024 00:30:00 -0400

This week’s guest is Jeff Kramer, VP Technology & Digital Factory at Kason Industries.

Jeff discusses why manufacturers struggle to develop a cohesive digital strategy, and lays out best practices around governance, data architecture, and bridging the IT/OT divide. He also explains why it’s critical for organizations to empower their frontline personnel by using technology to enable a citizen developer approach.

Episode 138: Democratizing Computer Vision with LandingAI’s Kai Yang

Tulip — Wed, 29 May 2024 00:15:00 -0400

This week’s guest is Kai Yang, VP of Product at LandingAI.

Kai discusses the need for a data-centric approach to AI, why vendors should build tools rather than solutions, and more, sharing lessons learned from his career in machine learning and software development. He also explains how new tools like visual prompting are democratizing computer vision and enabling anyone, regardless of skill level, to develop their own machine learning models.

Episode 137: AI for the Frontline Engineer with Instrumental’s Anna Shedletsky

Tulip — Wed, 15 May 2024 00:30:00 -0400

This week’s guest is Anna Shedletsky, Co-Founder and CEO of Instrumental.

Having started her career as an engineer at Apple, Anna shares lessons around quality management, which manufacturing KPIs actually matter, and how to take an idea from prototype to production. Plus, she lays out why organizations should think about manufacturing as a profit generator rather than a cost center, and why being able to demonstrate ROI is vital for engineers to advocate for the tech they need.

Episode 136: AI Takes Center Stage at Hannover Messe

Tulip — Wed, 01 May 2024 07:30:00 -0400

This week, Natan Linder, Co-Founder and CEO of Tulip sits down with Madilynn Castillo, Head of Marketing for a recap of their experiences at this year’s Hannover Messe — the world’s largest industrial trade fair.

They explore the lackluster state of AI in industrial software, the rise of composable software architectures, and how open technology ecosystems are becoming the norm throughout the industry. Plus, an overview of the latest developments from Tulip that debuted at the show.

Episode 135: Bringing Spatial Intelligence to Operations with Zerokey's Matt Lowe

Tulip — Wed, 10 Apr 2024 00:15:00 -0400

This week’s guest is Matt Lowe, Co-Founder and CEO of ZeroKey.

Not only is he a contributor to major open source projects like Linux and Arduino, Matt is the inventor of Quantum RTLS, a system that uses ultrasound to achieve 3D position tracking of objects with an unmatched level of fidelity. He explains what makes ultrasound-based positioning systems ideal for manufacturing environments, how spatial intelligence offers new ways to solve problems on the shop floor, and how open architecture can eliminate the need for system integrators.

Episode 134: Building Industrial Architectures with MQTT with HiveMQ’s Dominik Obermaier

Tulip — Wed, 27 Mar 2024 00:30:00 -0400

This week’s guest is Dominik Obermaier, Co-Founder and CTO of HiveMQ.

With over 10 years of experience serving on the MQTT technical committee and helping organizations build their data foundations using HiveMQ’s MQTT platform, Dominik shares his deep expertise on the technology. He explains what makes MQTT such an important communications protocol, why the emergence of the Unified Namespace matters for manufacturers, and debates the merits of on-prem vs. cloud solutions.

Episode 133: Rethinking Our Approach to AI with Dr. Jay Lee

Tulip — Wed, 13 Mar 2024 00:15:00 -0400

This week’s guest is Jay Lee, Director of the Industrial AI Center at the University of Maryland.

Dr. Lee shares his experiences from the early days programming machines with punch cards, to eventually developing advanced machine learning applications for industry. He explains how AI and ML are reshaping manufacturing, the workforce, and global supply chains. Plus, he lays out his vision for how our education system needs to change in order to train the next generation of AI practitioners.

Episode 132: Open Source Software for Manufacturing with UMH's Alex Krüger

Tulip — Wed, 28 Feb 2024 00:30:00 -0500

This week’s guest is Alex Krüger, Co-founder and CEO of United Manufacturing Hub, or UMH.

Alex shares his journey from working on integration projects in consulting fresh out of college, to founding UMH and building an open source alternative to the offerings from incumbent vendors. He breaks down the role of the open source software movement in manufacturing, how the Unified Namespace architecture compares to the traditional ISA-95 model, and how IT can best enable OT to solve problems. Plus, he shares his vision for how microservice-based MES solutions can disrupt the existing monolithic applications.

Episode 131: MQTT, Unified Namespace, and The New Industrial Data Stack with Litmus’s Vatsal Shah

Tulip — Wed, 14 Feb 2024 00:30:00 -0500

This week’s guest is Vatsal Shah, Founder and CEO of Litmus.

Vatsal discusses his journey from an automation engineer at Rockwell, to building a new industrial data platform from the ground up after becoming frustrated with the limitations of the offerings from established vendors. He discusses manufacturers’ exodus from on-prem to cloud systems, the pros and cons of data protocols like MQTT and Sparkplug B, and why the Unified Namespace architecture is getting so much attention. Plus, he shares his vision for the future of edge computing and how an open ecosystem of interoperable tools is transforming the industry.

Episode 130: Democratization, Gen AI, and the Future of Industrial Analytics with Seeq’s Lisa Graham

Tulip — Wed, 31 Jan 2024 00:30:00 -0500

This week’s guest is Dr. Lisa Graham, CEO of Seeq.

Dr. Graham discusses her journey from process engineer, to using Seeq’s platform as a customer, and now leading the company as CEO. Drawing on her extensive experience in operations, she discusses how advanced analytics, generative AI, and the emergence of an interoperable technology ecosystem are reshaping industries. Plus, she shares best practices for IT/OT collaboration, her vision for the future of historians, and how the democratization of data science is paving the way for a more efficient and sustainable future in operations and manufacturing.

Episode 129: AI and the Human Element in Industry 4.0 with Jeff Winter

Tulip — Wed, 17 Jan 2024 00:30:00 -0500

This week’s guest is Jeff Winter, Sr. Director of Industry Strategy for Manufacturing at Hitachi Solutions.

Jeff offers his insights into the history of the Industry 4.0 movement and how he expects it to evolve in the coming years. His discussion highlights the balance between AI and human ingenuity, the role of frontline workers in an increasingly automated manufacturing environment, and the untapped potential of manufacturing data.

Episode 128: From Tailor to Technologist: A Digital Transformation Journey with Joachim Hensch

Tulip — Wed, 03 Jan 2024 00:15:00 -0500

This week’s guest is Joachim Hensch, Founder of Joachim Hensch Consulting and former Managing Director of the Hugo Boss factory in Izmir, Turkey.

Hensch shares invaluable lessons learned about digital transformation through his over three decades of experience working in the apparel industry in roles from the shop floor all the way to management. His unique journey from tailor to digital transformation leader illustrates the realities of implementing Industry 4.0, challenges in traditional manufacturing, and the pressing need to empower workers with digital tools. Joachim discusses how manufacturers can balance artisanship with mass production by adopting new tools while retaining a deep appreciation of the frontline operators and their critical role in the industry.

Episode 127: Venture Capital's Role in Digital Transformation with Lior Susan

Tulip — Wed, 06 Dec 2023 00:15:00 -0500

This week’s guest is Lior Susan, founder of Eclipse Ventures.

With the digital transformation of critical industries like manufacturing now at the forefront of many nations’ economic priorities, Lior discusses the role that venture capital can play in helping drive this change. He addresses the growing importance of integrating IT and OT in industrial settings, and how technology can be used to augment the global workforce. Plus, key insights on the future of system integration in a world of open, interoperable software ecosystems.

Episode 126: Transforming Manufacturers’ Organizational Strategy with Dr. Jörg Gnamm

Tulip — Wed, 15 Nov 2023 00:15:00 -0500

This week’s guest is Dr. Jörg Gnamm, Senior Partner & Global Head of Manufacturing and Industry 4.0 Practice at Bain & Company.

In order to successfully transform their business, Jörg calls on manufacturers to take a systemic approach to technology adoption by enabling interdisciplinary collaboration, and focusing on use cases that drive value for the business. He draws on his extensive experience with real-world implementation examples, sharing his lessons learned and best practices from successfully implementing the blueprint he describes in our conversation.

Episode 125: Rethinking Quality Control for Pharmaceuticals with Mark Buswell

Tulip — Wed, 01 Nov 2023 00:15:00 -0400

Our guest this week is GSK’s Mark Buswell, VP of Quality Tech. Mark draws on over two decades of experience in pharma manufacturing as we explore the challenges of quality control in the industry.

Our discussion sheds light on the hurdles of adopting emerging technologies in regulated industries as Mark presents his vision of how to enable 'Quality by Design' with new tech and methodologies. He explains what the future of quality labs will look like, and what manufacturers need to do to prepare for this coming paradigm shift in the pharma industry.

Episode 124: Industrial Data Interoperability with Erich Barnstedt

Tulip — Wed, 18 Oct 2023 00:15:00 -0400

Our guest this week is Microsoft’s Erich Barnstedt, Chief Architect Standards, Consortia & Industrial IoT, Azure Edge + Platform.

Erich brings his perspective as we try to get to the bottom of why–despite overtures from some of the biggest vendors in the space–we still have not achieved true data interoperability in the manufacturing industry. We explore what really goes on behind the curtain at standards committees, and why it is so important for vendors to embrace an open technology ecosystem that puts interoperability at the forefront.

Episode 108: Lean Operations with John Carrier

Tulip — Wed, 15 Feb 2023 00:00:00 -0500

Augmented reveals the stories behind the new era of industrial operations, where technology will restore the agility of frontline workers.

In this episode of the podcast, the topic is "Lean Operations." Our guest is John Carrier, Senior Lecturer of Systems Dynamics at MIT. In this conversation, we talk about the people dynamics that block efficiency in industrial organizations.

If you like this show, subscribe at augmentedpodcast.co. Augmented is a podcast for industry leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

Follow the podcast on Twitter or LinkedIn.

Trond's Takeaway:

The core innovative potential in most organizations remains its people. The people dynamics that block efficiency can be addressed once you know what they are. But there is a hidden factory underneath the factory, which you cannot observe unless you spend time on the floor. And only with this understanding will tech investment and implementation really work. Stabilizing a factory is about simplifying things. That's not always what technology does, although it has the potential if implemented the right way.

Transcript:

TROND: Welcome to another episode of the Augmented Podcast. Augmented brings industrial conversations that matter, serving up the most relevant conversations on industrial tech. And our vision is a world where technology will restore the agility of frontline workers.

In this episode of the podcast, the topic is Lean Operations. Our guest is John Carrier, Senior Lecturer of Systems Dynamics at MIT. In this conversation, we talk about the people dynamics that block efficiency in industrial organizations.

Augmented is a podcast for industrial leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

John, welcome to the show. How are you?

JOHN: Trond, I'm great. And thank you for having me today.

TROND: So we're going to talk about lean operations, which is very different from a lot of things that people imagine around factories. John, you're an engineer, right?

JOHN: I am an engineer, a control engineer by training.

TROND: I saw Michigan in there, your way to MIT and chemical engineering, especially focused on systems dynamics and control. And you also got yourself an MBA. So you have a dual, if not a three-part, perspective on this problem. But tell me a little bit about your background. I've encountered several people here on this podcast, and they talk about growing up in Michigan. I don't think that's a coincidence.

JOHN: Okay, it's not. So I was born and raised in the city of Detroit. We moved out of the city, the deal of oil embargo in 1973. I've had a lot of relatives who grow up and work in the auto industry. So if you grew up in that area, you're just immersed in that culture. And you're also aware of the massive quote, unquote, "business cycles" that companies go through.

What I learned after coming to MIT and having the chance to meet the great Jay Forrester a lot of those business cycles are self-inflicted. What I do is I see a lot of the things that went right and went wrong for the auto industry, and I can help bring that perspective to other companies. [laughs]

TROND: And people have a bunch of assumptions about, I guess, assembly lines in factories. One thing is if you grew up in Michigan, it would seem to me, from previous guests, that you actually have a pretty clear idea of what did go on when you grew up in assembly lines because a lot of people, their parents, were working in manufacturing. They had this conception. Could we start just there? What's going on at assembly lines?

JOHN: I'm going to actually go back to 1975 to a Carrier family picnic. My cousin, who's ten years older than I, his summer job he worked at basically Ford Wayne, one of the assembly plants. He was making $12 an hour in 1975, so he paid his whole college tuition in like a month. But the interesting point was he was talking about his job when all the adults were around, and he goes, "Do you know that when they scratch the paint on the car, they let it go all the way to the end, and they don't fix it till it gets to the parking lot?"

And I'll never forget this. All the adults jumped on him. They're like, "Are you an idiot? Do you know how much it costs to shut the line down?" And if you use finance, that's actually the right answer. You don't stop the line because of a scratch; you fix it later. Keep the line running. It's $10,000 a minute. But actually, in the short term, that's the right decision. In the long term, if you keep doing that, you're building a system that simply makes defects at the same rate it makes product. And it's that type of logic and culture that actually was deeply ingrained in the thinking. And it's something that the Japanese car companies got away from.

It's funny how deeply ingrained that concept of don't stop the line is. And if you do that, you'll make defects at the same rate that you make product. And then, if you look at the Detroit newspapers even today, you'll see billion-dollar recalls every three months. And that's a cycle you've got to get yourself out of.

TROND: You know, it's interesting that we went straight there because it's, I guess, such a truism that the manufacturing assembly line kind of began in Detroit, or at least that's where the lore is. And then you're saying there was something kind of wrong with it from the beginning. What is it that caused this particular fix on keeping everything humming as opposed to, I guess, what we're going to talk about, which is fixing the system around it?

JOHN: There's a lot of work on this. There's my own perspective. There's what I've read. I've talked to people. The best I can come up with is it's the metrics that you pick for your company. So if you think about...the American auto industry basically grew up in a boom time, so every car you made, you made profit on. And their competitive metric was for General Motors to be the number one car company in the world.

And so what that means is you never miss a sale, so we don't have time to stop to fix the problem. We're just going to keep cranking out cars, and we'll fix it later. If you look at the Japanese auto industry, when it arose after World War II, they were under extreme parts shortages. So if one thing were broken or missing, they had to stop. So part of what was built into their culture is make it right the first time. Make a profit on every vehicle versus dominant market share.

TROND: Got it. So this, I guess, obsession with system that you have and that you got, I guess, through your education at MIT and other places, what is it that that does to your perspective on the assembly line? But there were obviously reasons why the Ford or the Detroit assembly lines, like you said, looked like they did, and they prioritized perhaps sales over other things.

When you study systems like this, manufacturing systems, to be very specific, how did you even get to your first grasp of that topic? Because a system, you know, by its very nature, you're talking about complexity. How do you even study a system in the abstract? Because that's very different, I guess, from going into an assembly and trying to fix a system.

JOHN: So it's a great question. And just one thing I want to note for the audience is although we talk about assembly lines, most manufacturing work is actually problem-solving and not simply repetitive. So we need to start changing that mindset about what operations really is in the U.S. We can come to that in the end.

TROND: Yeah.

JOHN: I'll tell you, I'm a chemical engineer. Three pieces of advice from a chemical engineer, the first one is never let things stop flowing. And the reason why that's the case in a chemical plant is because if something stops flowing for a minute or two, you'll start to drop things out of solution, and it will gum everything up. You'll reduce the capacity of that system till your next turnaround at least. And what happens you start getting sludge and gunk.

And for every class I was ever in, in chemical engineering, you take classes in heat transfer, thermodynamics, kinetics. I never took a class in sludge, [laughs] or sticky solids, or leftover inventory and blending. And then, when I first went to a real factory after doing my graduate work, I spent four to six years studying Laplace transforms and dynamics. All I saw were people running around. I'm like, that's not in the Laplace table.

And, again, to understand a chemical plant or a refinery, it takes you three to five years. So the question is, how can you actually start making improvement in a week when these systems are so complex? And it's watch the people running around. So that's why I focus a lot on maintenance teams. And I also work with operations when these things called workarounds that grow into hidden factories. So the magic of what I've learned through system dynamics is 80% to 90% of the time, the system's working okay, 10% or 20% it's in this abnormal condition, which is unplanned, unscheduled. I can help with that right away.

TROND: So you mentioned the term hidden factories. Can you enlighten me on how that term came about, what it really means? And in your practical work and consulting work helping people at factories, and operations teams, and maintenance teams, as you said, why is that term relevant, and what does it really do?

JOHN: Great. So I'm going to bring up the origin. So many people on this call recognize the name Armand Feigenbaum because when he was a graduate student at the Sloan School back in the '50s, he was working on a book which has now become like the bible, Total Quality Management or TQM. He's well known for that. He's not as well known for the second concept, which he should be better known for. Right after he graduated, he took a job in Pittsfield, Massachusetts, for one of the GE plastic plants.

Here he comes out of MIT. I'm going to apply linear equations. I'm going to do solving, all these mathematics, operation constraints, all these things. When he gets into that system, he realizes 30% of everything going on is unplanned, unscheduled, chaotic, not repeated. He's like, my mathematical tools just break down here.

So he did something...as important as marketing was as an operational objective, he named these things called hidden factories. And he said, 30% of all that work is in these hidden factories. And it's just dealing with small, little defects that we never ever solve. But over time, they actually erode our productivity of systems that can eat up 10% to 20% of productivity. And then, finally, it's work that I'm doing. It's the precursor to a major accident or disaster. And the good side is if you leave the way the system works alone, the 80%, and just focus on understanding and reducing these hidden factories, you can see a dramatic improvement quickly and only focus on what you need to fix.

TROND: So, for you, you focus on when the system falls apart. So you have the risk angle to this problem.

JOHN: Exactly. And so just two things, I'm like a doctor, and I do diagnosis. So when you go to the doctor, I'm not there to look at your whole system and fix everything. I'm like, here are first three things we got to work at, and, by the way, I use data to do that. And what I realized is if everyone just steps back after this call and thinks about today, right? When you get to the end of the day, what percent of everything in that factory or system happened that was in your schedule?

And you'll start to realize that 30% of the people are chasing symptoms. So you need data to get to that root cause, and that will tell you what data to collect. And second, look for time because what you're doing is these hidden factories are trying to keep the system running because you have a customer. You have your takt time, and so people are scrambling. And if you put that time back into the system, that's going to turn into product.

TROND: John, I'm just curious; when you say data, I mean, there's so much talk of data and big data and all kinds of data. But in manufacturing, apart from the parts that you're producing, I mean, some of this data is hard to come by. When you say data, what data will you even get access to?

JOHN: I come from the Albert Einstein School is. I need a ruler, and I need a stopwatch. Go into any system that you work in, whether it be your factory or your house, and ask the last time someone measured how long something took, and you will find a dearth of that data. And the reason why I love time data is it never lies. Most data I see in databases was collected under some context; I can't use it. So I go right in the floor and start watching 5 or 10 observations and looking at all the variation.

The second point I ask is, what's a minute worth in your system or a second? So if we're in an auto assembly plant, in a chemical plant, if we're in a hospital, in an operating room, those minutes and seconds are hundreds of thousands of dollars. So within about 20 minutes, not only have I measured where there's opportunity, we're already on the way to solving it.

TROND: So, so far, you haven't talked much about the technology aspects. So you work at a business school, but that business school is at MIT. There's a lot of technology there. It strikes me that a lot of times when we talk about improvements, certainly when we talk about efficiencies in factories, people bring up automation machines as the solution to that tool. And I'm sure you're not against machines, but you seem to focus a lot more on time, on organizational factors. How should people think about the technology factor inside of their operations?

JOHN: So, first, you brought up...my nickname is Dr. Don't. And the reason they call me Dr. Don't [laughs] is because they'll go, "Should we invest in this? Can we buy these robots?" I say, "No, you can't do that." And I'm going to tell you why. First is, I was quote, unquote, "fortunate enough" to work in a lot of small and mid-sized machine shops during the 2009 downturn. And I was brought in by the banks because they were in financial trouble.

And the one thing I noticed there was always a million-dollar automation or robot wrapped in plastic. And large companies can get away with overspending on technology, small and mid-sized companies can't. And so what you really want to do is go and watch and see what the problem is, buy just as much technology as you need, and then scale that.

First is, like I just said, I was just in a plant a few weeks ago, and they just implemented several hundred sensors to basically listen to their system. That's all good. It's data we need. Two problems, why'd you put in several hundred and not put in 20? And second, when we inspected it, about 15% were either not plugged in or weren't reading. So what happened was if we would have started with 20 and put the resource in analyzing that data, then when we scaled to the several hundred, we'd have had our systems in place. Instead, we overwhelmed everyone with data, so it really didn't change the way they work. Now we fixed that.

But your question was, why am I skeptical or slow to invest in technology? Technology costs money, and it takes time. If you don't look at the system first and apply the technology to solve the system problem, you're going to end up with a million-dollar piece of equipment wrapped in plastic. If you go the other direction, you will scale successfully. And no one's better at this than Toyota. They only invest in the technology they need. Yet you can argue they're at least as technologically sophisticated as all the rest. And they've never lost money except in 2009 so that is a proof point.

TROND: What are some examples of places you've been in lately, I don't know, individual names of companies? But you said you're working kind of mid-sized companies. Those are...[laughs] the manufacturing sector is mid-sized companies, so that sounds very relevant. But what are some examples in some industries where you have gone in and done this kind of work?

JOHN: I work for large companies and small and mid-sized. And I'm a chemical engineer, but I love machine shops. So I sit on the board of a $25 million machine shop. They make parts for a diesel truck and some military applications. They make flywheels. So one of their big challenges is in the United States and in the world, we're suffering with a problem with castings. We received our castings. Interesting thing is there are void fractions.

One of the things I do want to share is as a systems guy, I'm not an expert in mechanical engineering or any of that, but I can add value by helping look for defects. Let me tell you what their challenge is. So, first of all, more of their castings are bad. Then this surprised me...I learned from asking questions. If you've ever been in a machine shop, one thing I learned about when you're making casting is that there are always bubbles in it. You can't avoid it.

The art of it is can you put the bubbles in the places where they don't hurt? You minimize the bubbles, and you move them to the center. So one is we're getting bad castings, but the second part was when we made some of these castings, and they had a void problem in the center. So that doesn't cause a problem with your flywheel. The customer sent them back because they're becoming oversensitive to the defects that don't count. And it's because they switched out staff.

So I guess what I'm trying to say here is our supply chain is undergoing this new type of stress because we're losing the type of expert system expertise that we've had from people that have worked in this industry 20 to 30 years. That's a really important aspect.

The second is we're in their line balancing all the time. I think a lot of the things you learn in class, you spend one class on load balancing or line balancing, operation and manufacturing, and then you go into a factory, and no one's doing it. So I just wanted to share two points. My one factor is doing that they cut 30% of their time.

Another system I'm working in they have one experienced supervisor managing four new people on four different setups. What I realized is there's not enough of that supervisor to go around. We're like, why don't we shoot videos like the NFL does [laughs] and watch those films of how people do their work? Because when you're an expert, Trond, and you go to do a task, you say, "That has five steps."

But if I sent you or me new, we'd look and go, "There are really about 80 steps in there." And you explained it to me in 15 minutes. How am I going to remember that? So shooting film so people can go back and watch instead of bothering your supervisor all the time, which they won't do. So what I do think, to wrap up on this point, is when you talk about technology, the camera, the video that you have in your pocket, or you can buy for $200, is the best technology you can probably apply in the next three to six months. And I would greatly encourage everyone to do something like that.

MID-ROLL AD:

In the new book from Wiley, Augmented Lean: A Human-Centric Framework for Managing Frontline Operations, serial startup founder Dr. Natan Linder and futurist podcaster Dr. Trond Arne Undheim deliver an urgent and incisive exploration of when, how, and why to augment your workforce with technology, and how to do it in a way that scales, maintains innovation, and allows the organization to thrive. The key thing is to prioritize humans over machines.

Here's what Klaus Schwab, Executive Chairman of the World Economic Forum, says about the book: "Augmented Lean is an important puzzle piece in the fourth industrial revolution."

Find out more on www.augmentedlean.com, and pick up the book in a bookstore near you.

TROND: I wanted to ask you then, derived from this, to what extent can some of these things be taught as skills on a systemic level in a university or in some sort of course, and to what extent? Do you really just have to be working in manufacturing and observing and learning with data on your own? By extension, to what extent can a manager or someone, anyone in the organization, just develop these practices on their own? And to what extent do you need mentorship from the outside to make it happen or see something in the system that is very difficult to see from the inside?

JOHN: So it's interesting you ask that because that's very much the problem I'm dealing with because as good as our universities are, the best place to learn operations in manufacturing is on the factory floor. So how do you simulate that approach? I teach lean operations at MIT Sloan. And what I do with my students is I ask them to pick a routine task, video two minutes of it, and reduce that by 30%. And I've done this two years in a row.

When you look at these projects, the quality of the value streams and the aha moments they had of time that they were losing is stunning. You know what the challenge is? They don't yet always appreciate how valuable that is. And what I want them to realize is if you're washing dishes or running a dishwasher, why is that any different from running a sterilization process for hospital equipment? Why is that any different from when you're actually doing setup so that maintenance can get their work done 30% faster?

I've given them the tools, and hopefully, that will click when they get out into the workspace. But I do have one success point. I had the students...for some classes, they have to run computers and simulations during class. So that means everyone has to have the program set up. They have to have the documentation. So you can imagine 5 to 10 minutes a class, people getting everything working right.

One of my teams basically said we're going to read...it took about five minutes, and they said, we're going to do this in 30 seconds just by writing some automated scripts. They did that for our statistics class, and then they shared it with their other classmates, beautiful value stream, video-d the screens, did it in about four or five hours. The next class they took later I found out they did that for a class project, and they sold the rights to a startup. So first is getting them that example in their own space, and then two, helping them make analogies that improving things in your own house isn't all that much different than the systemic things in a factory.

TROND: Learning by analogy, I love it. I wanted to profit from your experience here on a broader question. It takes a little bit more into the futuristic perspective. But in our pre-conversation, you talked about your notion on industry 4.0, which, to me, it's a very sort of technology, deterministic, certainly tech-heavy perspective anyway.

But you talked about how that for you is related to..., and you used another metaphor and analogy of a global nervous system. What do you think, well, either industry 4.0 or the changes that we're seeing in the industry having to do with new approaches, some of them technology, what is it that we're actually doing with that? And why did you call it a global nervous system?

JOHN: When I graduated from school, and I'm a control systems skilled in the arts, so to speak. And the first thing I did...this is back in the '90s, so we're industry 3.0. When you're in a plant, no one told me I was going to spend most of my time with the I&C or the instrumentation and control techs and engineers. That's because getting a sensor was unbelievably expensive. Two, actually, even harder than getting the budget for it was actually getting the I&C tech's time to actually wire it up. It would take six weeks to get a sensor.

And then three, if it weren't constantly calibrated and taken care of, it would fall apart. And four, you get all those three workings, if no one's collecting or knows how to analyze the data, you're just wasting [laughs] all your money. So what was exciting to me about industry 4.0 was, one, the cost of sensors has dropped precipitously, two, they're wireless with magnets. [laughs] So the time to set it up is literally minutes or hours rather than months and years.

Three, now you can run online algorithms and stuff, so, basically, always check the health of these sensors and also collect the data in the form. So I can go in, and in minutes, I can analyze what happened versus, oh, I got to get to the end of the week. I never looked at that sensor. And four, what excited me most, and this gets to this nervous system, is if you look at the way industries evolved, what always amazes me is we got gigantic boilers and train engines and just massive equipment, physical goods. Yet moving electrons actually turns out to be much more costly in the measurement than actually building the physical device.

So we're just catching up on our nervous system for the factory. If I want to draw an analogy, if you think about leprosy; a lot of people think leprosy is a physical disease; what it is is it's your nerves are damaged, so because your nerves are damaged, you overuse that equipment, and then you wear off your fingers. And if you look at most maintenance problems in factories, it's because they didn't have a good nervous system to realize we're hurting our equipment.

And maintenance people can't go back and say, "Hey, in three months, you're going to ruin this." And the reason I know it is because I have this nervous system because I'm measuring how much you're damaging it rather than just waving it. And now it becomes global because, let's say you and I have three pumps in our plant, and we need to take care of those. They are on the production line, very common. What if we looked at the name of that pump, called the manufacturer who's made tens of thousands of those? There's the global part.

So they can help us interpret that data and help us take care of it. So there's no defect or failure that someone on this planet hasn't seen. It's just we never had the ability to connect with them and send them the data on a platform like we can with a $5,000 pump today. So that's why I look at it, and it's really becoming a global diagnosis.

TROND: It's interesting; I mean, you oscillate between these machine shops, and you had a medical example, but you're in medical settings as well and applying your knowledge there. What is the commonality, I guess, in this activity between machine shops, you know, improving machine shops and improving medical teams' ability to treat disease and operate faster? What is it that is the commonality?

So you've talked about the importance, obviously, of communication and gathering data quicker, so these sensors, obviously, are helping out here. But there's a physical aspect. And, in my head, a machine shop is quite different from an operating room, for example. But I guess the third factor would be human beings, right?

JOHN: I'm going to put an analogy in between the machine shops at the hospital, and that's an F1 pit crew. And the reason I love F1 is it's the only sport where the maintenance people are front and center. So let's now jump to hospitals, so the first thing is if I work in a hospital, I'm talking to doctors or nurses in the medical community. And I start talking about saving time and all that. Hey, we don't make Model Ts. Every scenario we do is different, and we need to put the right amount of time into that surgery, which I completely agree to.

Where we can fix is, did we prepare properly? Are all our toolkits here? Is our staff trained and ready? And you'd think that all those things are worked out. I want to give two examples, one is from the literature, and one is from my own experience. I'd recommend everyone look up California infant mortality rates and crash carts. The state of California basically, by building crash carts for pregnancies and births, cut their infant mortality rate by half just by having that kit ready, complete F1 analogy. I don't want my surgeon walking out to grab a knife [laughs] during surgery.

And then second is, I ran a course with my colleagues at MIT for the local hospitals here in Boston. You know what one of the doctor teams did over the weekend? They built one of these based on our class. They actually built...this is the kit we want. And I was unbelievably surprised how when we used the F1 analogy, the doctors and surgeons loved it, not because we're trying to actually cut their time off. We're trying to put the time into the surgery room by doing better preparations and things like that. So grabbing the right analogy is key, and if you grab the right analogy, these systems lessons work across basically anywhere where time gets extremely valuable.

TROND: As we're rounding off, I wanted to just ask you and come back to the topic of lean. And you, you use the term, and you teach a class on lean operations. Some people, well, I mean, lean means many things. It means something to, you know, in one avenue, I hear this, and then I hear that.

But to what extent would you say that the fundamental aspects of lean that were practiced by Toyota and perhaps still are practiced by Toyota and the focus on waste and efficiency aspects to what extent are those completely still relevant? And what other sort of new complements would you say are perhaps needed to take the factory to the future, to take operational teams in any sector into their most optimal state?

JOHN: As a control engineer, I learned about the Toyota Production System after I was trained as a control system engineer. And I was amazed by the genius of these people because they have fundamentally deep control concepts in what they do. So you hear concepts like, you know, synchronization, observability, continuous improvement. If you have an appreciation for the deep control concepts, you'll realize that those are principles that will never die.

And then you can see, oh, short, fast, negative feedback loops. I want accurate measurements. I always want to be improving my system. With my control background, you can see that this applies to basically any system. So, in fact, I want to make this argument is a lot of people want to go to technology and AI. I think the dominant paradigm for any system is adaptive control. That's a set of timeless principles.

Now, in order to do adaptive control, you need certain technologies that provide you precision analysis, precision measurement, real-time feedback loops. And also, let us include people into the equation, which is how do I train people to do tasks that are highly variable that aren't applying automation is really important. So I think if people understand, start using this paradigm of an adaptive control loop, they'll see that these concepts of lean and the Toyota Production System are not only timeless, but it's easier to explain it to people outside of those industries.

TROND: Are there any lessons finally to learn the way that, I guess, manufacturing and the automotive sector has been called the industry of industries, and people were very inspired by it in other sectors and have been. And then there has been a period where people were saying or have been saying, "Oh, maybe the IT industry is more fascinating," or "The results, you know, certainly the innovations are more exciting there." Are we now at a point where we're coming full circle where there are things to learn again from manufacturing, for example, for knowledge workers?

JOHN: What's driving the whole, whether it be knowledge work or working in a factory...which working in a factory is 50% knowledge work. Just keep that in mind because you're problem-solving. And you know what's driving all this? It is the customer keeps changing their demands. So for a typical shoe, it'll have a few thousand skews for that year. So the reason why manufacturing operations and knowledge work never get stale is the customer needs always keep changing, so that's one.

And I'd like to just end this with a comment from my colleague, Art Byrne. He wrote The Lean Turnaround Action Guide as well as has a history back to the early '80s. And I have him come teach in my course. At his time at Danaher, which was really one of the first U.S. companies to successfully bring in lean and Japanese techniques, they bring in the new students, and the first thing they put them on was six months of operations, then they move to strategy and finance, and all those things.

The first thing that students want to do is let's get through these operations because we want to do strategy and finance and all the marketing, all the important stuff. Then he's basically found that when they come to the end of the six months, those same students are like, "Can we stay another couple of months? We just want to finish this off." I'm just saying I work in the floor because it's the most fun place to work.

And if you have some of these lean skills and know how to use them, you can start contributing to that team quickly. That's what makes it fun. But ultimately, that's why I do it. And I encourage, before people think about it, actually go see what goes on in a factory or system before you start listening to judgments of people who, well, quite frankly, haven't ever done it. So let me just leave it at that. [laughs]

TROND: I got it. I got it. Thank you, John. Spend some time on the floor; that's good advice. Thank you so much. It's been very instructive. I love it. Thank you.

JOHN: My pleasure, Trond, and thanks to everybody.

TROND: You have just listened to another episode of the Augmented Podcast with host Trond Arne Undheim. The topic was Lean operations, and our guest was John Carrier, Senior Lecturer of Systems Dynamics at MIT. In this conversation, we talked about the people dynamics that block efficiency in industrial organizations.

My takeaway is that the core innovative potential in most organizations remains its people. The people dynamics that block efficiency can be addressed once you know what they are. But there is a hidden factory underneath the factory, which you cannot observe unless you spend time on the floor. And only with this understanding will tech investment and implementation really work. Stabilizing a factory is about simplifying things. That's not always what technology does, although it has the potential if implemented the right way.

Thanks for listening. If you liked the show, subscribe at augmentedpodcast.co or in your preferred podcast player, and rate us with five stars. If you liked this episode, you might also like other episodes on the lean topic. Hopefully, you'll find something awesome in these or in other episodes, and if so, do let us know by messaging us. We would love to share your thoughts with other listeners.

The Augmented Podcast is created in association with Tulip, the frontline operation platform that connects people, machines, and devices, and systems. Tulip is democratizing technology and empowering those closest to operations to solve problems. Tulip is also hiring, and you can find Tulip at tulip.co.

Please share this show with colleagues who care about where industrial tech is heading. And to find us on social media is easy; we are Augmented Pod on LinkedIn and Twitter and Augmented Podcast on Facebook and YouTube.

Augmented — industrial conversations that matter. See you next time. Special Guest: John Carrier.

Episode 107: Explainability in AI with Julian Senoner

Tulip — Wed, 01 Feb 2023 00:00:00 -0500

Augmented reveals the stories behind the new era of industrial operations, where technology will restore the agility of frontline workers.

In this episode of the podcast, the topic is "Explainability and AI." Our guest is Julian Senoner, CEO and Co-Founder of EthonAI. In this conversation, we talk about how to define explainable AI and its major applications, and its future.

If you like this show, subscribe at augmentedpodcast.co. If you like this episode, you might also like Episode 103: Human-First AI with Christopher Nguyen.

Augmented is a podcast for industry leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

Follow the podcast on Twitter or LinkedIn.

Trond's Takeaway:

Explainability in AI, meaning knowing exactly what's going on with an algorithm, is very important for industry because its outputs must be understandable to the process engineers using it. The computer has not and will not use the product. Only a domain expert can recognize when the system is wrong, and that will be the case for a very long time in most production environments.

Transcript:

TROND: Welcome to another episode of the Augmented Podcast. Augmented reveals the stories behind a new era of industrial operations where technology will restore the agility of frontline workers. Technology is changing rapidly. What's next in the digital factory, and who's leading the change?

In this episode of the podcast, the topic is Explainability and AI. Our guest is Julian Senoner, CEO and Co-Founder of EthonAI. In this conversation, we talk about how to define explainable AI and its major applications, and its future.

Augmented is a podcast for industrial leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

Julian, welcome to the show.

JULIAN: Hello, Trond. Thank you for having me.

TROND: I'm excited to have you. You know, you're a fellow runner; that's always good. And you grew up in the ski slopes.; that makes me feel at home as a Norwegian. So you grew up in Austria; that must have been pretty exciting. And then you were something as exciting as a ski instructor in the Alps. That's every man and woman's dream.

JULIAN: Yeah, I think it was very nice to grow up in the mountains. I enjoyed it a lot. But, you know, times have passed, and now I'm happy to be in Zurich.

TROND: You went on to industrial engineering. You studied manufacturing and production at ETH. And you got interested in statistics and machine learning aspects of all of that. How did this happen? You went from ski instruction to statistics.

JULIAN: Yeah, I was always impressed about watching stuff being made. I think it's a very relaxing thing to do. And I always wanted to become an engineer. When I was five years old, I wanted to become a ship engineer. So it was always clear that I wanted to do something with manufacturing and mechanical engineering. So I started actually doing my bachelor's in Vienna at Technische University. And for my master's, I moved to Zurich and studied Industrial Engineering.

ETH has historically been very strong in machine learning research. Every student, no matter if you're interested or not, gets exposed to machine learning, statistics, and AI. It caught my attention. I thought there were very interesting things you can do when you combine both. So that's how I ended up doing research on interface and becoming an entrepreneur in this area.

TROND: Yeah, we'll talk about your entrepreneurship in a moment. But I wanted to go to your dissertation, Artificial Intelligence in Manufacturing: Augmenting Humans at Work. That is very close to our interests here at the podcast. Tell me more about this.

JULIAN: There is a lot of hype about AI. There's a lot of talk about self-aware factories and these kinds of things. These predictions are not new. We had studies in the 1970s that predicted there won't be any people in factories by the 1980s; everything will be run by a centralized computer. I never believed in these kinds of things. During my dissertation, I was interested in looking into how we can develop useful AI tools that can support people doing their jobs more effectively and efficiently.

TROND: Right. But you already were onto this idea of humans at work. Where did you do your case studies? I understand ABB and Siemens were two of them. Give me a little sense of what you discovered there; pick any one of them.

JULIAN: Sure. I'd love to start with the case that we ran with Siemens. I've worked quite a lot with Siemens in different use cases, but one of them was supporting frontline workers in complex assembly tasks on electronic products. So the aim was to help the worker check if the product has been assembled correctly. There are many connectors that could be missing or assembled in the wrong way. So the idea was to have a camera mounted to the workstation, and the worker would put the final product under the camera and get visual feedback if it has been assembled correctly or not.

What we did here is really studying the psychological aspect of that. I would say most of my Ph.D. was really math-heavy and about modeling, but here we were interested in the psychological aspect. Because, in the beginning, we thought perhaps andon lights with green or red signals would be enough. But we got intrigued by the research question of does the worker actually follow these recommendations if it's just the green or red light?

So we did an experiment, which I'm very excited about. So we got 50 workers that volunteered within Siemens to participate, which I'm very grateful for. We basically divided factory workers in two groups. We looked into the effect of explainability in the decisions that the AI makes. So we had one group that got basically just a recommendation to reject or to pass the product. And we had another group that got the exact same recommendations. But in addition to that, we provided visual feedback indicating the area where the AI believes that there could be an error.

And the results of this experiment were perhaps not too surprising, but the effect size clearly was. We found that the people that did not get explanations for these recommendations were more than three times more likely to overrule the AI system, although the AI was correct. And I think this is a really nice finding.

TROND: Well, it's super interesting in terms of trust in AI. And this topic of explainability is so much talked about these days, I guess, not always in manufacturing because people overlook that sometimes, people who are not in the industry. And they think about whether machines will take over and what decisions they're taking over and, certainly, if the machines are part of the decision making, what goes into that decision making?

But as you were discovering more about explainability, what is explainability? And how is it different from even just being able to...it starts, I guess, with the decision of the AI being very clear because if that's not even clear, then you can't even interpret the decision. But then there's a lot of discussion in the industry, I mean, in the AI field, I guess, about interpretability. So can you actually understand the process?

But you did this experiment, and it became very clear, it seems, that just the decision is not enough. Was it the visual example that was helping here? Or what is it that people want to know about a machine decision to make them trust the decision and trust that their processes, you know, remains a good process?

JULIAN: I think I kind of see two answers to this question; one is the aspects of interpretability and explainability; perhaps I start with that. So these terms are often used interchangeably, and academics are still arguing about the differences. But there is now a popular opinion that I also share that these two things are not the same.

So when we talk about interpretable AI, we think about models that have basically an interpretable architecture or functional relationship, so an example would be a linear regression. You have a regression line; it has a slope, it has an intercept. And you know how an X translates into a Y or an input to an output.

Explainable AI is a fairly new research branch, which it's slightly different to that. It looks into more complex AI models like deep neural networks and ensembling techniques, which do not have this inherently interpretable model architecture. So a human, just by looking at it, cannot understand how decisions are formed. And what explainable AI methods do is basically reverse engineer what the model is doing by approximating the inner behavior of the model. So, in essence, we're creating a model of a model.

Coming to your second question, so why might this be important in manufacturing? Basically, what I discovered during my research is that AI is still not trusted in the manufacturing domain, so people often do not understand what AI does in general, and I think explanations are a very powerful tool to simplify that. And a second use case of explainability is also that we can reduce complexity. So we can use more powerful methods to model more complex relationships. And we can use explainability on top of that to, for example, conduct problem-solving.

TROND: Wow, you explain it very easily, but it's not easy to explain an actual AI model. Like, if you were to say, you know, here is the neural network model I used, and it had eight layers, good luck explaining that to a manufacturing worker or to me.

JULIAN: So I think that explaining what a model is is also a different topic, and perhaps it's not even needed. You can still treat this as a mathematical function. I think it's really more about the decisions. We need to understand how decisions are formed, and there are different techniques to that.

So when we talk about vision models, heat maps are a very interesting application. So we cannot really tell how the algorithm came up with a decision, but we can try to visualize the areas of an image where the neural network focused on to inform its decision. And we can, for example, see that certain areas are more highlighted than others, and perhaps that goes with the human intuition and creates more trust.

TROND: You know, this topic is, for me, so fascinating because when we think about frontline workers or, indeed, engineers or quality managers on the shop floor, previously, they didn't have perhaps the tools available to open up the boxes, to open up the machines and look at the decisions that were being made. And, of course, that doesn't lead you to an enormous amount of confidence that what you're doing is good or bad or mediocre. You're not getting enough feedback.

But it does seem to me that as machines and tools and algorithms on the shop floor become more and more complex, this is going to be a big effort. It doesn't sound very easy. And maybe you can characterize, you know, with the process today. These methods are just being applied on the shop floor. Do you have a sense that this general idea that things have to be explainable is a shared commodity on shop floors that are starting to use these techniques? Or would you say that it's enough of a challenge just to start experimenting with them, let alone trying to explain them to anybody around? I guess it's called a black box problem, right?

JULIAN: Sure. I think in any use case where you have some kind of interaction between humans and AI, you've got to have explainability. It's going to be key. And there are also some less obvious use cases around explainable AI that I would also be happy to share. I think everywhere where you're going towards full automation, you perhaps don't need it, perhaps only if it's very high-stakes decisions being made. If you are rejecting products based on an AI that are very expensive, you might want to know why your line scrapped the products. In general, if you're going for automation, I would say explainability is nice to have. When we talk about augmentation, I think it's absolutely key.

TROND: Yeah, it's absolutely clear. But we were talking before, and you were reminding me that in semiconductor production on an average production line, a large percentage of those components tend to be scrapped for quality reasons. And each of those components might be very, very expensive to manufacture. And it's a big problem.

You have to recycle the part again, and they're made out of rare earth metals or whatnot. And it's a complicated thing, so it's not like you're just making wooden parts that you can just do over or plastic, and you can mold it again. These are like you said, they're expensive decisions that you're trusting machines to make.

JULIAN: Yes. Talking about semiconductor industry, we have been also working on a different use case using explainable AI in semiconductors which I'm really excited about, and that is root cause analysis. In semiconductor manufacturing, as you mentioned, it's common that manufacturers throw away 15% to 20% of the chips they produce. We have car production lines that are standing still because of a chip shortage, so, obviously, this is a problem.

What explainable AI can do here is we can try to model relationships in the manufacturing system to try to understand what causes these problems in the first place. So actually, when I was still a researcher back at ETH, I worked together with ABB semiconductors who exactly had this problem, so they had costly yield losses. And process engineers were struggling to find out where these losses were coming from.

Because in semiconductor manufacturing, you often have hundreds of process steps, and each process, you can have even hundreds of process parameters such as temperatures and pressures. And you would like to know which of these process parameters do I need to adjust to avoid my yield losses. And if you have thousands, in this case, we had 3,600 different parameters that could have been suspects for yield losses. It's very hard to kind of track where yield losses are coming from.

And the methods that are still used in industry are often based on linear methods. So we find this big effect, but we can't find the more hidden ones. And I think this is a very neat application of AI because you can use more complex models like neural networks or tree-based methods to model these relationships. So, in essence, we try to imitate the physical processes to learn the physical processes as they are.

But since neural networks are very complex and we cannot really understand what they have been modeling, we need kind of explanations for that. And using these explanations, we can inform process engineers who are domain experts about what the model might have found. We're still acting upon correlations, not causation. But still, we can point towards certain areas that are interesting.

And in the case of ABB, the AI guided the process engineers to suspicious processes, and the domain experts were able to come up with two improvement actions based on that input. Then they were able to reduce the scrap by more than 50% in one of their lines, which was, of course, substantial. And I think it's a very nice example of how humans and AI can collaborate and get more out of it.

TROND: So, Julian, is that what augmentation of workers means to you? The augmented workforce is essentially a collaboration between man and machine in a deeper way than before?

JULIAN: I think it's about expanding capabilities and getting teams of humans and machines that perform better and provide more value than either alone. I think that is what augmentation is about.

MID-ROLL AD:

Here's what Klaus Schwab, Executive Chairman of the World Economic Forum, says about the book: "Augmented Lean is an important puzzle piece in the fourth industrial revolution."

Find out more on www.augmentedlean.com, and pick up the book in a bookstore near you.

TROND: You ended up commercializing an idea around this. Tell me more about that and how that came about.

JULIAN: Exactly. So at some point in my Ph.D., I decided I'm not going to become a professor; many PhDs have a similar experience. But I saw that 50% scrap reduction, for example, at one partner company that's quite substantial, and these tools kind of scale. And together with a friend of mine who did a Ph.D. in the same department, we decided let's found a company around this. So we founded a startup called EthonAI.

And we are developing a software platform that helps manufacturers improve their quality management. So we offer five different products in different application areas, three computer vision products that help manufacturers in detecting defects, one product around process monitoring and anomaly detection, and one product for AI-based root cause analysis. So we cover this entire continuous improvement loop that manufacturers need, and we provide tools for that.

And I think one thing that really stands out for us, and this is our key hypothesis, is that all of our products can be used by process engineers without writing a single line of code. We're not building fancy toolboxes for data scientists. We really build the tools for the people in the factory. We want to empower them to do their jobs better or help them do it better because we believe these are the people that drive the improvements on the shop floor. And those people should be the ones that use the tools.

TROND: I'm just curious; in this process, you've talked to a lot of process engineers. How are they reacting to these new opportunities? Are they excited that they get to do some programming or do advanced analysis without getting deeply into software? Or are they a little bit weary that they have to still jump into a new domain? There's so much discussion these days about the need for re-skilling. How have you found them to react to these new changes?

JULIAN: I think the experience is really mixed. Sometimes you see skepticism; sometimes, you see great excitement. I think at the end of the day, you need to solve problems. It's not about bringing AI to the factory. It's rather to solve the problems of the people that have worked in those factories. And if you have a tool that can be used very easily...so anyone that can operate a computer can operate our tools, and it solves problems. People are often happy to use it or mostly happy to use it.

Of course, if you are coming in with this marketing thing of AI is going to change everything, then you experience more skepticism. So this is why we really talk about problems and not about the technology. Technology kind of is in the back end. And people don't care about how fancy your algorithms are; it has to work. What I think is so rewarding working on a startup in the manufacturing space is that the outcomes are so binary; it either works or it doesn't. And that's really cool to see in the physical world. Our entire team is working really hard to bring very useful tools to the market.

TROND: You're a little bit of a hybrid yourself between a manufacturing engineer, basically, and now a little bit of a software engineer or at least an analytics perspective here with statistics and machine learning. Where do you find the expertise that you need at Ethon to move forward?

Because it's a very rare thing still to combine knowledge of shop floor real-world challenges, systems that cannot fail, or at least when they fail, it has a bigger consequence than when a software needs to patch because the whole idea of why your software tentatively is in the shop floor is to reduce these kinds of stops and starts at production lines. How do you find people that really managed to combine the perspective of building software with this reality of how it's going to work in a physical environment?

JULIAN: We have hired three kinds of people, first being brilliant scientists from the area of AI that have never been in a factory before but have been publishing at the highest level, NeurIPS, ICML, all those top outlets in the AI field. And they can really help to push the state of the art.

The second category of people is process engineers. We've hired process engineers from companies that could be potential clients. They know about the problems and have been conducting all this data analysis in a quite manual way. So they're coming into our company to kind of guide us in building the product of their dreams. Then we have some people that have been working on the interface, such as Bernhard and myself. So Bernhard is the CTO in our company, people that have been on the interface between AI and manufacturing.

So I think you really need basically experts and generalists in a company, and then it typically goes well. And then the other thing that you need is really customer-centricity. So you really need to be close to your customers and try to understand their pains. And you also need to bring the machine learning engineers to the factories so that they can see for themselves, and that's usually very helpful.

TROND: What's your experience with frontline operators themselves? In the experiments that you've done with your Ph.D. and with your early products with Ethon, how are they reacting to these new opportunities for augmentation?

JULIAN: I mean, typically, they're quite positive about it because it helps them in their jobs. But when we're talking about augmentation, it's not about replacing people; it's about helping them do things better. And often, they're quite critical about user interfaces, but you learn so much interacting with them, and you improve it. And then they get something that they really need.

And the biggest learnings here have really been around how do you visualize or represent things or content to the workers? This is so crucial because, in factories, you have such a diverse workforce of people in their 60s who will retire soon. And you have young people who just started their first job and basically were raised with iPads. So it's really that's a challenge, but it's a great one to work on.

TROND: Well, I guess that brings me to the important question that a lot of people want to answer about the factory of the future. There's fear about the factory of the future. There's this idea of a 24/7 automated factory, autonomous, working without humans. And then there's this idea that factories will look very different and people will do different things, but there are still people in there.

And then there's anything in between. You could also imagine a world without factories but where industrial production happens somewhat distributed. Because I guess what a factory is, at the heart of it, is it's a mass of things, of people and machines concentrated in one location.

JULIAN: Yes, it's a social, technical system. As I mentioned in the beginning, there is a lot of hype about AI, and these predictions are not new. This lights-out factory has been around for a very long time, and I still haven't seen it, at least on a productive level. You can also over-automate. Companies have seen that Tesla, for example, has cut back on automation. Elon Musk tweeted, "Humans are underrated."

So I think in the future, people will still shape the image of factories. Of course, we will see more automation that is enabled through AI because, with AI, we can also now model stochastic processes. So we don't need deterministic outcomes like a robot that always picks the same things. We can have more self-organized, isolated systems. But I think it's really about isolated systems and not an entire factory that is operated by one single artificial brain.

But I think the most intriguing use cases will be those where we augment humans rather than replacing them. So I think that's really where the magic happens. It's giving process engineers, for example, the tools to conduct more effective problem solving, finding things that were unknown to them previously, and couple that with their domain expertise. I think that's certainly something that we're going to see more.

I don't think that AI will regulate production processes in a fully automated manner. At the end of the day, you always need the process and product expert that has intuition about the processes, creative problem-solving. So I think process engineers will always be around and be integral to any manufacturing system, but I think AI we will see more AI tools that enable these people and empower them. And I think this is an exciting development.

TROND: Yeah, it's so interesting that sometimes I feel like these futuristic discussions fail to take into account innovation. So it's a very basic problem with this discussion because you're assuming that automation and factory production overall is all about squeezing out tiny, little efficiencies, and if that's all it is, then machines might be the better way to go.

But if you're talking about incrementally improving and sometimes radically improving a process or changing even what you are producing based on feedback from a market and stuff like that, it would seem to me that we are quite far into the future before a socio-technical system like that with complete feedback, long supply chain, and understanding what all of these things mean and the decisions that go into it, and costs. And that all can be managed by one network algorithm. It's really a little bit hard to envision how those futurists really have been thinking about it, or maybe they were just considering isolated use of robots that looked very cool.

JULIAN: The thing with AI is that AI is incredibly lazy; that's one of the problems. It always tries to learn shortcuts. And we need to understand the things that it learns. You always have the example of correlation and causation. And I think if you provide outputs from an AI to a human expert who can judge the validity of the results, that's where you can generate value. If an AI is supposed to make fully automated decisions, then I think it would relatively quickly turn out to be a mess because it might just be a correlation and not a causal relationship.

There's always this famous example of shark attacks and ice cream consumption, both have a very high correlation, but it's not because sharks like ice cream. So I think it's very similar in manufacturing. When you produce stuff, you might find a correlation with a certain process parameter that can't causally have an effect on your quality, for example, or downtime. And if you would have a system that operates itself, it very likely would try to tweak that parameter with perhaps a bad outcome. So I think the human in the loop still remains crucial here.

TROND: What excites you about the future of manufacturing? Everyone's always worried about manufacturing because it's such a big part of the economy. A lot of people lose their jobs if things go badly. And, historically, it's gone up and down. And maybe for a while, in some countries, it's gone mostly down, and it hasn't been the most exciting place to be. Are you excited about the future of manufacturing?

JULIAN: Yes, I mean, definitely. I've always been, and I think, I will always be excited about manufacturing. And obviously, at EthonAI, we will also try to leave a mark on the industry. We are helping big companies to improve everything around quality but also helping them improve their CO2 footprint by reducing production waste. And this is something that really excites me to help these companies to provide useful tools and also see that these tools have an impact in the physical world at big corporations like Siemens. That's a very exciting place to be. I think we will see very, very interesting developments over the next couple of years.

TROND: Well, Julian, it's exciting to jump in and hear a little bit about your world here. I certainly wish you best of luck with Ethon, and it was fascinating to hear. Thank you so much for sharing your perspective.

JULIAN: Thank you very much, Trond.

TROND: You have just listened to another episode of the Augmented Podcast with host Trond Arne Undheim. The topic was Explainability in AI. Our guest was Julian Senoner, CEO and Co-Founder of EthonAI. In this conversation, we talked about how to define explainable AI, its major applications, and its future.

My takeaway is that explainability in AI, meaning knowing exactly what's going on with an algorithm, is very important for industry because its outputs must be understandable to the process engineers using it. The computer has not and will not use the product. Only a domain expert can recognize when the system is wrong, and that will be the case for a very long time in most production environments.

Thanks for listening. If you liked the show, subscribe at augmentedpodcast.co or in your preferred podcast player, and rate us with five stars. If you liked this episode, you might also like Episode 103: Human-First AI with Christopher Nguyen. Hopefully, you'll find something awesome in these or in other episodes, and if so, do let us know.

The Augmented Podcast is created in association with Tulip, the frontline operation platform that connects people, machines, devices, and systems. Tulip is democratizing technology and empowering those closest to operations to solve problems. Tulip is also hiring, and you can find Tulip at tulip.co.

Please share this show with colleagues who care about where industrial tech is heading. You can find us on social media, and we are Augmented Pod on LinkedIn and Twitter and Augmented Podcast on Facebook and YouTube.

Augmented — industrial conversations that matter. See you next time. Special Guest: Julian Senoner.

Episode 103: Human-First AI with Christopher Nguyen

Tulip — Wed, 23 Nov 2022 00:00:00 -0500

Augmented reveals the stories behind the new era of industrial operations, where technology will restore the agility of frontline workers.

In this episode of the podcast, the topic is Human-First AI. Our guest is Christopher Nguyen, CEO, and Co-Founder of Aitomatic. In this conversation, we talk about the why and the how of human-first AI because it seems that digital AI is one thing, but physical AI is a whole other ballgame in terms of finding enough high-quality data to label the data correctly. The fix is to use AI to augment existing workflows. We talk about fishermen at Furuno, human operators in battery factories at Panasonic, and energy optimization at Westinghouse.

If you like this show, subscribe at augmentedpodcast.co. If you like this episode, you might also like Episode 80: The Augmenting Power of Operational Data, with Tulip's CTO, Rony Kubat.

Augmented is a podcast for industry leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

Follow the podcast on Twitter or LinkedIn.

Trond's Takeaway:

Physical AI is much more interesting of a challenge than pure digital AI. Imagine making true improvements to the way workers accomplish their work, helping them be better, faster, and more accurate. This is the way technology is supposed to work, augmenting humans, not replacing them. In manufacturing, we need all the human workers we can find. As for what happens after the year 2100, I agree that we may have to model what that looks like. But AIs might be even more deeply embedded in the process, that's for sure.

Transcript:

TROND: Welcome to another episode of the Augmented Podcast. Augmented brings industrial conversations that matter, serving up the most relevant conversations in industrial tech. Our vision is a world where technology will restore the agility of frontline workers.

Augmented is a podcast for industrial leaders, process engineers, and for shop floor operators hosted by futurist Trond Arne Undheim and presented by Tulip.

Christopher, how are you? And welcome.

CHRISTOPHER: Hi, Trond. How are you?

TROND: I'm doing great. I thought we would jump into a pretty important subject here on human-first AI, which seems like a juxtaposition of two contradictory terms, but it might be one of the most important types of conversations that we are having these days.

I wanted to introduce you quickly before we jump into this. So here's what I've understood, and you correct me if I'm wrong, but you are originally from Vietnam. This is back in the late '70s that you then arrived in the U.S. and have spent many years in Silicon Valley mostly. Berkeley, undergrad engineering, computer science, and then Stanford Ph.D. in electrical engineering. You're a sort of a combination, I guess, of a hacker, professor, builder. Fairly typical up until this point of a very successful, accomplished sort of Silicon Valley immigrant entrepreneur, I would say, and technologist.

And then I guess Google Apps is something to point out. You were one of the first engineering directors and were part of Gmail, and Calendar, and a bunch of different apps there. But now you are the CEO and co-founder of Aitomatic. What we are here to talk about is, I guess, what you have learned even in just the last five years, which I'm thrilled to hear about. But let me ask you this first, what is the most formational and formative experience that you've had in these years? So obviously, immigrant background and then a lot of years in Silicon Valley, what does that give us?

CHRISTOPHER: I guess I can draw from a lot of events. I've always had mentors. I can point out phases of my life and one particular name that was my mentor. But I guess in my formative years, I was kind of unlucky to be a refugee but then lucky to then end up in Silicon Valley at the very beginning of the PC revolution. And my first PC was a TI-99/4A that basically the whole household could afford. And I picked it up, and I have not stopped hacking ever since. So I've been at this for a very long time.

TROND: So you've been at this, which is good because actually, good hacking turns out takes a while. But there's more than that, right? So the story of the last five years that's interesting to me because a lot of people learn or at least think they learn most things early. And you're saying you have learned some really fundamental things in the last five years. And this has to do with Silicon Valley and its potential blindness to certain things. Can you line that up for us? What is it that Silicon Valley does really well, and what is it that you have discovered that might be an opportunity to improve upon?

CHRISTOPHER: Well, I learn new things every four or five years. I actually like to say that every four or five years, I look back, and I say, "I was so stupid five years ago." [laughs] So that's been the case.

TROND: That's a very humbling but perhaps a very smart knowledge acquisition strategy, right?

CHRISTOPHER: Yeah. And in the most recent five years...so before co-founding Aitomatic, which is my latest project and really with the same team...and I can talk a lot more about that. We've worked with each other for about ten years now. But in the intervening time, there's a four-and-a-half-year block when we were part of Panasonic. So we had a company called Arimo that was acquired by Panasonic for our machine learning AI skills and software.

And I would say if you look at my entire history, even though I did start with my degree in semiconductor all the way down to device physics and Intel and so on, but in terms of a professional working career, that was the first time we actually faced the physical world as a Silicon Valley team. And anybody who's observed Silicon Valley in the last 15-20 years, certainly ten years, has seen a marked change in terms of the shift from hardware to software. And my friend Marc Andreessen likes to say, "Software is eating the world."

If you look at education, you know, the degrees people are getting, it has shifted entirely from engineering all the way to computer science. And the punch line, I guess, the observation is that we Silicon Valley people do not get physical. We don't understand the manufacturing world. We don't know how to do HVAC and so on. And so when we build software, we tend to go for the digital stuff.

TROND: Christopher, it's almost surprising given the initial thrust of Silicon Valley was, of course, hardware. So it's not surprising to me, I guess because I've been observing it as well. But it is striking more than surprising that a region goes through paradigms.

CHRISTOPHER: Yeah. Yeah. And it's a global trend. It's the offshoring of low-end, shall we say, low-value manufacturing and so on. And we're discovering that we actually went a little too far. So we don't have the skill set, the expertise anymore. And it's become a geopolitical risk.

TROND: Right. Well, a little bit too far, maybe, or not far enough. Or, I mean, tell us what it is that you're losing when you lose the hardware perspective, particularly in this day and age with the opportunities that we're about to talk about.

CHRISTOPHER: Well, I can talk specifically about the things that touch my immediate spheres. Maybe you can think abstractly about the lack of tooling expertise and manufacturing know-how, and so on. But as part of Panasonic, the acquisition was all about taking a Silicon Valley team and injecting AI, machine learning across the enterprise. And so we were part of that global AI team reporting to the CTO office.

And we found out very quickly that a lot of the software techniques, the machine learning, for example, when you think about people saying data is the fuel for machine learning and specifically labeled data, right? In the digital world, the Google place that I came from, it was very easy to launch a digital experiment and collect labels, decisions made by users. You can launch that in the morning, and by evening you're building examples. You can't do that in the physical world. Atoms move a lot more slowly. And so when you try to do something like predictive maintenance, you don't have enough failure examples to train machine learning models from.

So all of the techniques, all of the algorithms that we say we developed from machine learning that seem to work so well, it turns out it worked so well because the problem space that we worked on has been entirely digital, and they all fail when it comes to manufacturing, the things that you can touch and feel, you know, cars that move and so on.

TROND: I want to ask you this, Christopher, because the first company you helped co-found was, in fact, a contract manufacturer. Do you think that reflecting on this long career of yours and these various experiences, what was it that convinced you before others? I mean, you're not the only one now in the Valley that has started to focus on manufacturing and including hardware again, but it is rare still. What does it require to not just think about manufacturing but actually start to do compute for manufacturing? Is it just a matter of coming up with techniques? Or is it a whole kind of awareness that takes longer? So, in your case, you've been aware of manufacturing, acutely aware of it for decades.

CHRISTOPHER: I would say there are two things, one is obvious, and the other was actually surprising to me. The obvious one is, of course, knowledge and experience. When we work on sonar technology that shoots a beam down an echogram that comes back to detect fish in the ocean, it's very necessary, not just convenient, but necessary for the engineers that work on that to understand the physics of sound waves travel underwater, and so on.

So that education, I have long debates, and it's not just recently. When we were trying to structure a syllabus for a new university, I had long debates with my machine-learning friends, and they said, "We don't need physics." And I said, "We need physics." That's one thing. But you can concretely identify you need to know this. You need to know this. So if you're going to do this, learn the following thing.

The thing that was more unexpected for me in the last five years as I sort of sound this bell of saying, hey, we need to modify our approach; we need to optimize our algorithms for this world, is a cultural barrier. It's kind of like the story of if you have a hammer, you want to go look for nails. So Silicon Valley today does not want to look for screwdrivers yet for this world.

TROND: So you're saying Silicon Valley has kind of canceled the physical world? If you want to be really sort of parabolic about this, it's like software is eating the world, meaning software is what counts, and it's so efficient. Why go outside this paradigm, basically? If there's a problem that apparently can't be fixed by software, it's not a valuable problem.

CHRISTOPHER: Or I can't solve that problem with my current approach. I just have to squint at it the right way. I have to tweak the problem this way and so on despite the fact that it's sort of an insurmountable challenge if you tried to do so. And concretely, it is like, just give me enough data, and I'll solve it. And if you don't have enough data, you know what? Go back and get more data. [chuckles] That's what I myself literally said. But people don't have the luxury of going back to get more data. They have to go to market in six months, and so on.

TROND: Right. And so manufacturing...and I can think of many use cases where obviously failure, for example, is not something...you don't really want to go looking for more failure than you have or artificially create failure in order to stress test something unless that's a very safe way of doing so. So predictive maintenance then seems like a, I guess, a little bit of a safer space. But what is it about that particular problem that then lends itself to this other approach to automating labeling? Or what exactly is it that you are advocating one should do to bridge to digital and the physical AIs?

CHRISTOPHER: I actually disagree that it is a safer space.

TROND: Oh, it's not a safer space to you.

CHRISTOPHER: That itself there's a story in that, so let's break that down.

TROND: Let's do it.

CHRISTOPHER: So, again, when I say Silicon Valley, it is a symbol for a larger ecosystem that is primarily software and digital. And when I say we, because I've worn many hats, I have multiple wes, including academia; I've been a professor as well. When we approach the predictive maintenance problem, if you approach it as machine learning, you got to say, "Do this with machine learning," the first thing you ask for...let's say I'm a data scientist; I'm an AI engineer.

You have this physical problem. It doesn't matter what it is; just give me the dataset. And the data set must have rows and columns, and the rows are all the input variables. And then there should be some kind of column label. And in this case, it'll be a history of failures of compressors failing, you know, if the variables are such, then it must be a compressor. If the variables are such, it must be the air filter, and so on.

And it turns out when you ask for that kind of data, you get ten rows. [laughs] That's not enough to do machine learning on. So then people, you know, machine learning folks who say they've done predictive maintenance, they actually have not done predictive maintenance. That's the twist. What they have done is anomaly detection, which machine learning can do because, with anomaly detection, I do not need that failure label. It just gives me all the sensor data.

What anomaly detection really does is it learns the normal patterns. If you give it a year's worth of data, it'll say, okay, now I've seen a year's worth of data. If something comes along that is different from the past patterns; I will tell you that it's different. That's only halfway to predictive maintenance. That is detecting that something is different today. That is very different from, and it isn't predicting, hey, that compressor is likely to fail about a month from now.

And that when we were part of Panasonic, it turns out the first way...and we solved it exactly the way I've described. We did it with the anomaly detection. And then we threw it over the wall to the engineer experts and said, "Well, now that you have this alert, go figure out what may be wrong." And half of the time, they came back and said, "Oh, come on, it was just a maintenance event. Why are you bothering me with this?"

TROND: But, Christopher, leveraging human domain expertise sounds like a great idea. But it can't possibly be as scalable as just leveraging software. So how do you work with that? And what are the gains that you're making?

CHRISTOPHER: I can show you the messenger exchange I had with another machine-learning friend of mine who said exactly the same thing yesterday, less than 24 hours ago.

TROND: [laughs]

CHRISTOPHER: He said, "That's too labor-intensive." And I can show you the screen.

TROND: And how do you disprove this?

CHRISTOPHER: Well, [chuckles] it's not so much disproving, but the assumption that involving humans is labor-intensive is only true if you can't automate it. So the key is to figure out a way, and 10-20 years ago, there was limited technology to automate or extract human knowledge, expert systems, and so on. But today, technologies...the understanding of natural language and so on, machine learning itself has enabled that. That turns out to be the easier problem to solve. So you take that new tool, and you apply it to this harder physical problem.

TROND: So let's go to a hard, physical problem. You and I talked about this earlier, and let's share it with people. So I was out fishing in Norway this summer. And I, unfortunately, didn't get very much fish, which obviously was disappointing on many levels. And I was a little surprised, I guess, of the lack of fish, perhaps. But I was using sonar to at least identify different areas where people had claimed that there were various types of fish. But I wasn't, I guess, using it in a very advanced way, and we weren't trained there in the boat.

So we sort of had some sensors, but we were not approaching it the right way. So that helped me...and I know you work with Furuno, and Garmin is the other obviously player in this. So fish identification and detection through sonar technology is now the game, I guess, in fishery and, as it turns out, even for individuals trying to fish these days. What is that all about? And how can that be automated, and what are the processes that you've been able to put in place there?

CHRISTOPHER: By the way, that's a perfect segue into it. I can give a plug perhaps for this conference that I'm on the organizing committee called Knowledge-First World. And Furuno is going to be presenting their work exactly, talking a lot about what you're talking about. This is kind of coming up in November. It is the first conference of its kind because this is AI Silicon Valley meets the physical world.

I think you're talking about the fish-finding technology from companies like Furuno, and they're the world's largest market share in marine navigation and so on. And the human experts in this are actually not even the engineers that build these instruments; it's the fishermen, right? The fishermen who have been using this for a very long time combine it with their local knowledge, you know, warm water, cold water, time of day, and so on. And then, after a while, they recognize patterns that come back in this echogram that match mackerel, or tuna, or sardines, and so on.

And Furuno wants to capture that knowledge somehow and then put that model into the fish-finding machine that you and I would hold. And then, instead of seeing this jumbled mess of the echogram data, we would actually see a video of fish, for example. It's been transformed by this algorithm.

TROND: So, I mean, I do wish that we lived in a world where there was so much fish that we didn't have to do this. But I'm going to join your experiment here. And so what you're telling me is by working with these experts who are indeed fishermen, they're not experts in sonar, or they're not experts in any kind of engineering technology, those are obviously the labelers, but they are themselves giving the first solutions for how they are thinking about the ocean using these technologies. And then somehow, you are turning that into an automatable, an augmented solution, essentially, that then can find fish in the future without those fishermen somehow being involved the next time around because you're building a model around it.

CHRISTOPHER: I'll give you a concrete explanation, a simplified version of how it works, without talking about the more advanced techniques that are proprietary to Furuno. The conceptual approach is very, very easy to understand, and I'll talk about it from the machine learning perspective.

Let's say if I did have a million echograms, and each echogram, each of these things, even 100,000, is well-labeled. Somebody has painstakingly gone through the task of saying, okay, I'm going to circle this, and that is fish. And that is algae, and that's sand, and that's marble. And by the way, this is a fish, and this is mackerel, and so on. If somebody has gone through the trouble of doing that, then I can, from a human point of view, just run an algorithm and train it. And then it'll work for that particular region, for that particular time. Okay, well, we need to go collect more data, one for Japan, the North Coast, and one for Southwestern.

So that's kind of a lot of work to collect essentially what this pixel data is, this raw data. When you present it to an experienced fisherman, he or she would say, "Well, you see these bubbles here, these circles here with a squiggly line..." So they're describing it in terms of human concepts. And then, if you sit with them for a day or two, you begin to pick up these things. You don't need 100,000-pixel images. You need these conceptual descriptions.

TROND: So you're using the most advanced AI there is, which is the human being, and you're using them working with these sonar-type technologies. And you're able to extract very, very advanced models from it.

CHRISTOPHER: The key technology punch line here is if you have a model that understands the word circle and squiggly line, which we didn't before, but more recently, we begin to have models, you know, there are these advances called large language models. You may have heard of GPT-3 and DALL-E and so on, you know, some amazing demonstrations coming out of OpenAI and Google. In a very simplified way, we have models that understand the world now. They don't need raw pixels. These base models are trained from raw pixels, but then these larger models understand concepts. So then, we can give directions at this conceptual level so that they can train other models. That's sort of the magic trick.

TROND: So it's a magic trick, but it is still a difficult world, the world of manufacturing, because it is physical. Give me some other examples. So you worked with Panasonic. You're working with Furuno in marine navigation there and fishermen's knowledge. How does this work in other fields like robotics, or with car manufacturing, or indeed with Panasonic with kind of, I don't know, battery production or anything that they do with electronics?

CHRISTOPHER: So, to give you an example, you mentioned a few things that we worked on, you know, robotics in manufacturing, robotics arm, sort of the manufacturing side, and the consistency of battery sheets coming off the Panasonic manufacturing line in Sparks, Nevada as well as energy optimization at Westinghouse. They supply into data centers, and buildings, and so on.

And so again, in every one of these examples, you've got human expertise. And, of course, this is much more prevalent in Asia because Asia is still building things, but some of that is coming back to the U.S. There are usually a few experts. And by the way, this is not about thousands of manufacturing line personnel. This is about three or four experts that are available in the entire company. And they would be able to give heuristics. –They will be able to describe at the conceptual level how they make their decisions.

And if you have the technology to capture that in a very efficient way, again, coming back to the idea that if you make them do the work or if you automate their work, but in a very painstaking way like thousands of different rules, that's not a good proposition. But if you have some way to automate the automation, automate the capturing of that knowledge, you've got something that can bridge this physical, digital divide.

MID-ROLL AD:

Here's what Klaus Schwab, Executive Chairman of the World Economic Forum, says about the book: "Augmented Lean is an important puzzle piece in the fourth industrial revolution."

Find out more on www.augmentedlean.com, and pick up the book in a bookstore near you.

TROND: How stable is that kind of model knowledge? Because I'm just thinking about it in the long run here, are these physical domain experts that are giving up a little bit of their superpower are they still needed then in a future scenario when you do have such a model? Or will it never be as advanced as they are? Or is it actually going to be still kind of an interface that's going to jump between machines and human knowledge kind of in a continuous loop here?

CHRISTOPHER: Yeah, in the near term, it turns out we're not working on replacing experts as much as scaling experts. Almost every case we've worked on, companies are in trouble largely because the experts are very, very few and far between, and they're retiring. They're leaving. And that needs to be scaled somehow. In the case of, for example, the cold chain industry all of Japan servicing the supermarkets, you know, there's 7-ELEVEN, there's FamilyMart, and so on, there are three experts who can read the sensor data and infer what's likely to fail in the next month. So in the near term, it's really we need these humans, and we need more of them.

TROND: I'm glad to hear that even that is a bit of a contrarian message. So you're saying physical infrastructure and the physical world matters. You're saying humans matter. [laughs] It's interesting. Yeah, that's contrarian in Silicon Valley, I'll tell you that.

CHRISTOPHER: It is. And, in fact, related to that problem, Hussmann, which is a refrigeration company, commercial refrigeration supplies to supermarkets. It was a subsidiary of Panasonic. It has a really hard time getting enough service personnel, and they have to set up their own universities, if you will, to train them. And these are jobs that pay very well. But everybody wants to be in software these days.

Coming back to the human element, I think that long-term I'm an optimist, not a blind optimist but a rational one. I think we're still going to need humans to direct machines. The machine learning stuff is data that reflects the past, so patterns of the past, and you try to project that in the future. But we're always trying to effect some change to the status quo. Tomorrow should be a better day than today. So is that human intent that is still, at least at present, lacking in machines? And so we need humans to direct that.

TROND: So what is the tomorrow of manufacturing then? How fast are we going to get there? Because you're saying, well, Silicon Valley has a bit of a learning journey. But there is language model technology or progress in language models that now can be implemented in software and, through humans, can be useful in manufacturing already today. And they're scattered examples, and you're putting on an event to show this. What is the path forward here, and how long is this process? And will it be an exponential kind of situation here where you can truly integrate amazing levels of human insight into these machine models? Or will it take a while of tinkering before you're going to make any breakthroughs?

Because one thing is the breakthrough in understanding human language, but what you're saying here is even if you're working only with a few experts, you have to take domain by domain, I'm assuming, and build these models, like you said, painstakingly with each expert in each domain. And then, yes, you can put that picture together. But the question is, how complex of a picture is it that you need to put together? Is it like mapping the DNA, or is it bigger? Or what kind of a process are we looking at here?

CHRISTOPHER: If we look at it from the dimension of, say, knowledge-based automation, in a sense, it is a continuation. I believe everything is like an s-curve. So there's acceleration, and then there's maturity, and so on. But if you look back in the past, which is sort of instructive for the future, we've always had human knowledge-based automation.

I remember the first SMT, the Surface Mount Technology, SMT wave soldering machine back in the early '90s. That was a company that I helped co-found. It was about programming the positioning of these chips that would just come down onto the solder wave. And that was human knowledge for saying, move it up half a millimeter here and half a millimeter there. But of course, the instructions there are very micro and very specific.

What machine learning is doing...I don't mean to sort of bash machine learning too much. I'm just saying culturally, there's this new tool really that has come along, and we just need to apply the tool the right way. Machine learning itself is contributing to what I described earlier, that is, now, finally, machines can understand us at the conceptual level that they don't have to be so, so dumb as to say, move a millimeter here, and if you give them the wrong instruction, they'll do exactly that. But we can communicate with them in terms of circles and lines, and so on.

So the way I see it is that it's still a continuous line. But what we are able to automate, what we're able to ask our machines to do, is accelerating in terms of their understanding of these instructions. So if you can imagine what would happen when this becomes, let's say, ubiquitous, the ability to do this, and I see this happening over the next...Certainly, the base technology is already there, and the application always takes about a decade.

TROND: Well, the application takes a decade. But you told me earlier that humans should at least have this key role in this knowledge-first application approach until 2100, you said, just to throw out a number out there. That's, to some people, really far away. But the question is, what are you saying comes after that? I know you throw that number out.

But if you are going to make a distinction between a laborious process of painful progress that does progress, you know, in each individual context that you have applied to human and labeled it, and understood a little case, what are we looking at, whether it is 2100, 2075, or 2025? What will happen at that moment? And is it really a moment that you're talking about when machines suddenly will grasp something very, very generic, sort of the good old moment of singularity, or are you talking about something different?

CHRISTOPHER: Yeah, I certainly don't think it's a moment. And, again, the HP-11C has always calculated Pi far faster and with more digits than I have. So in that sense, in that particular narrow sense, it's always been more intelligent than I am.

TROND: Yeah. Well, no one was questioning whether a calculator could do better calculations than a human. For a long time --

CHRISTOPHER: Hang on. There's something more profound to think about because we keep saying, well, the minute we do something, it's okay; that's not intelligence. But what I'm getting to is the word that I would refer to is hyper-evolution. So there's not a replacement of humans by machines. There's always been augmentation, and intelligence is not going to be different. It is a little disturbing to think about for some of us, for a lot of us, but it's not any different from wearing my glasses.

Or I was taking a walk earlier this morning listening to your podcast, and I was thinking how a pair of shoes as an augmented device would seem very, very strange to humans living, say, 500 years ago, the pair of shoes that I was walking with. So I think in terms of augmenting human intelligence, there are companies that are working on plugging in to the degree that that seems natural or disturbing. It is inevitable.

TROND: Well, I mean, if you just think about the internet, which nowadays, it has become a trope to think about the internet. I mean, not enough people think about the internet as a revolutionary technology which it, of course, is and has been, but it is changing. But whether you're thinking about shoes, or the steam engine, or nuclear power, or whatever it is, the moment it's introduced, and people think they understand it, which most people don't, and few of us do, it seems trivial because it's there.

CHRISTOPHER: That's right.

TROND: But your point is until it's there, it's not trivial at all. And so the process that you've been describing might sound trivial, or it might sound complex, but the moment it's solved or is apparently solved to people, we all assume that was easy. So there's something unfair about how knowledge progresses, I guess.

CHRISTOPHER: That's right. That's right. We always think, yeah, this thing that you describe or I describe is very, very strange. And then it happens, and you say, "Of course, that's not that interesting. Tell me about the future."

TROND: Well, I guess the same thing has happened to cell phones. They were kind of a strange thing that some people were using. It was like, okay, well, how useful is it to talk to people without sitting by your desk or in the corner of your house?

CHRISTOPHER: I totally remember when we were saying, "Why the hell would I want to be disturbed every moment of the day?" [laughs] I don't want the phone with me, and now I --

TROND: Right. But then we went through the last decade or so where we were saying, "I can't believe my life before the phone." And then maybe now the last two, three years, I would say a lot of people I talk to or even my kids, they're like, "What's the big deal here? It's just a smartphone," because they live with a smartphone. And they've always had it.

CHRISTOPHER: They say, "How did you get around without Google Maps?" And then somebody says, "We used maps." And I said, "Before Google Maps."

[laughter]

TROND: Yeah. So I guess the future here is an elusive concept. But I just want to challenge you one more time then on manufacturing because manufacturing, for now, is a highly physical exercise. And, of course, there's virtual manufacturing as well, and it builds on a lot of these techniques and machine learning and other things. How do you see manufacturing as an industry evolve? Is it, like you said, for 75 years, it's going to be largely very recognizable? Is it going to look the same? Is it going to feel the same?

Is the management structure the way engineers are approaching it, and the way workers are working? Are we going to recognize all these things? Or is it going to be a little bit like the cell phone, and we're like, well, of course, it's different. But it's not that different, and it's not really a big deal to most people.

CHRISTOPHER: Did you say five years or 50 years?

TROND: Well, I mean, you give me the timeframe.

CHRISTOPHER: Well, in 5 years, we will definitely recognize it, but in 50 years, we will not

TROND: In 50 years, it's going to be completely different, look different, feel different; factories are all going to be different.

CHRISTOPHER: Right, right. I mean, the cliché is that we always overestimate what happens in 5 and underestimate what happens in 50. But the trend, though, is there's this recurring bundling and unbundling of industries; it's a cycle. Some people think it's just, you know, they live ten years, and they say it's a trend, but it actually goes back and forth. But they're sort of increasing specialization of expertise.

So, for example, the supply chain over the last 30 years, we got in trouble because of that because it has become so discrete if you want to use one friendly word, but you can also say fragmented in another word. Like, everybody has been focused on just one specialization, and then something like COVID happens and then oh my God, that was all built very precisely for a particular way of living. And nobody's in the office anymore, and we live at home, and that disrupts the supply chain.

I think if you project 50 years out, we will learn to essentially matrix the whole industry. You talked about the management of these things. The whole supply chain, from branding all the way down to raw materials, is it better to be completely vertically integrated to be part of this whole mesh network? I think the future is going to be far more distributed. But there'll be fits and starts.

TROND: So then my last question is, let's say I buy into that. Okay, let's talk about that for a second; the future is distributed or decentralized, whatever that means. Does that lessen or make globalization even more important and global standardization, I guess, across all geographical territories? I'm just trying to bring us back to where you started with, which was in the U.S., Silicon Valley optimized for software and started thinking that software was eating the world. But then, by outsourcing all of the manufacturing to Asia, it forgot some essential learning, which is that when manufacturing evolves, the next wave looks slightly different. And in order to learn that, you actually need to do it.

So does that lesson tell you anything about how the next wave of matrix or decentralization is going to occur? Is it going to be...so one thought would be that it is physically distributed, but a lot of the insights are still shared. So, in other words, you still need global insight sharing, and all of that is happening. If you don't have that, you're going to have pockets that are...they might be very decentralized and could even be super advanced, but they're not going to be the same. They're going to be different, and they're going to be different paths and trajectories in different parts of the world.

How do you see this? Do you think that our technology paradigms are necessarily converging along the path of some sort of global master technology and manufacturing? Or are we looking at scattered different pictures that are all decentralized, but yet, I don't know, from a bird's eye view, it kind of looks like a matrix?

CHRISTOPHER: I think your question is broader than just manufacturing, although manufacturing is a significant example of that, right?

TROND: It's maybe a key example and certainly under-communicated. And on this podcast, we want to emphasize manufacturing, but you're right, yes.

CHRISTOPHER: The word globalization is very loaded. There's the supposedly positive effect in the long run. But who is it that said...is it Keynes that said, "In the long run, we're all dead?" [laughs] In the short run, the dislocations are very real. A skill set of a single human being can't just shift from hardware to software, from manufacturing to AI, within a few months.

But I think your question is, let's take it seriously on a scale of, say, decades. I think about it in terms of value creation. There will always be some kind of disparity. Nature does not like uniformity. Uniformity is coldness; it is death. There have to be some gradients. You're very good at something; I'm very good at something else. And that happens at the scale of cities and nations as well.

TROND: And that's what triggers trade, too, right?

CHRISTOPHER: Exactly.

TROND: Because if we weren't different, then there would be no incentive to trade.

CHRISTOPHER: So when we think about manufacturing coming back to the U.S., and we can use the word...it is correct in one sense, but it's incorrect in another sense. We're not going back to manufacturing that I did. We're not going back to surface mount technology. In other words, the value creation...if we follow the trajectory of manufacturing alone and try to learn that history, what happens is that manufacturing has gotten better and better. Before, we were outsourcing the cheap stuff. We don't want to do that. But then that cheap stuff, you know, people over there build automation and skills, and so on. And so that becomes actually advanced technology.

So in a sense, what we're really doing is we're saying, hey, let's go advanced at this layer. I think it's going to be that give and take of where value creation takes place, of course, layered with geopolitical issues and so on.

TROND: I guess I'm just throwing in there the wedge that you don't really know beforehand. And it was Keynes, the economist, that said that the only thing that matters is the short term because, in the end, we are all dead eventually. But the point is you don't really know. Ultimately, what China learned from manufacturing pretty pedestrian stuff turned out to be really fundamental in the second wave.

So I'm just wondering, is it possible to preempt that because you say, oh, well, the U.S. is just going to manufacture advanced things, and then you pick a few things, and you start manufacturing them. But if you're missing part of the production process, what if that was the real advancement? I guess that is what happened.

CHRISTOPHER: Okay. So when I say that, I think about the example of my friend who spent, you know, again, we were a Ph.D. group at Stanford together. And whereas I went off to academia and did startups and so on, he stayed at Intel for like 32 years. He's one of the world's foremost experts in semiconductor process optimization. So that's another example where human expertise, even though semiconductor manufacturing is highly automated, you still need these experts to actually optimize these things. He's gone off to TSMC after three decades of being very happy at one place.

So what I'm getting to is it is actually knowable what are the secret recipes, where the choke points are, what matters, and so on. And interestingly, it does reside in the human brain. But when I say manufacturing coming back to the U.S. and advanced manufacturing, we are picking and choosing. We're doing battery manufacturing. We're doing semiconductor, and we're not doing wave soldering.

So I think it is possible to also see this trend that anybody who's done something and going through four or five iterations of that for a long time will become the world's expert at it. I think that is inevitable. You talk of construction, for example; interestingly, this company in Malaysia that is called Renong that is going throughout Southeast Asia; they are the construction company of the region because they've been doing it for so long. I think that is very, very predictable, but it does require the express investment in that direction. And that's something that Asia has done pretty well.

TROND: Well, these are fascinating things. We're not going to solve them all on this podcast. But definitely, becoming an expert in something is important, whether you're an individual, or a company, or a country for sure. What that means keeps changing. So just stay alert, and stay in touch with both AI and humans and manufacturing to boot. It's a mix of those three, I guess. In our conversation, that's the secret to unlocking parts of the future. Thank you, Christopher, for enlightening us on these matters. I appreciate it.

CHRISTOPHER: It's my pleasure.

TROND: You have just listened to another episode of the Augmented Podcast with host Trond Arne Undheim. The topic was Human-First AI. Our guest was Christopher Nguyen, CEO, and Co-Founder of Aitomatic. In this conversation, we talked about the why and the how of human-first AI because it seems that digital AI is one thing, but physical AI is a whole other ballgame.

My takeaway is that physical AI is much more interesting of a challenge than pure digital AI. Imagine making true improvements to the way workers accomplish their work, helping them be better, faster, and more accurate. This is the way technology is supposed to work, augmenting humans, not replacing them. In manufacturing, we need all the human workers we can find. As for what happens after the year 2100, I agree that we may have to model what that looks like. But AIs might be even more deeply embedded in the process, that's for sure.

Thanks for listening. If you liked the show, subscribe at augmentedpodcast.co or in your preferred podcast player, and rate us with five stars. If you liked this episode, you might also like Episode 80: The Augmenting Power of Operational Data, with Tulip's CTO, Rony Kubat as our guest. Hopefully, you'll find something awesome in these or in other episodes, and if so, do let us know by messaging us. We would love to share your thoughts with other listeners.

The augmented podcast is created in association with Tulip, the frontline operation platform that connects the people, machines, devices, and systems used in a production and logistics process in a physical location. Tulip is democratizing technology and empowering those closest to operations to solve problems. Tulip is also hiring. You can find Tulip at tulip.co.

Please share this show with colleagues who care about where industry and especially about how industrial tech is going. To find us on social media is easy; we are Augmented Pod on LinkedIn and Twitter and Augmented Podcast on Facebook and on YouTube.

Augmented — industrial conversations that matter. See you next time. Special Guest: Christopher Nguyen.

Episode 97: Industrial AI

Tulip — Wed, 21 Sep 2022 00:00:00 -0400

Augmented reveals the stories behind the new era of industrial operations, where technology will restore the agility of frontline workers.

The topic is Industrial AI. Our guest is Professor Jay Lee, the Ohio Eminent Scholar, the L.W. Scott Alter Chair Professor in Advanced Manufacturing, and the Founding Director of the Industrial AI Center at the University of Cincinnati.

In this conversation, we talk about how AI does many things but to be applicable; the industry needs it to work every time, which puts additional constraints on what can be done by when.

If you liked this show, subscribe at augmentedpodcast.co. If you liked this episode, you might also like Episode 81: From Predictive to Diagnostic Manufacturing Augmentation.

Augmented is a podcast for industry leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim and presented by Tulip.

Follow the podcast on Twitter or LinkedIn.

Trond's Takeaway:

Industrial AI is a breakthrough that will take a while to mature. It implies discipline, not just algorithms. In fact, it entails a systems architecture consisting of data, algorithm, platform, and operation.

Transcript:

TROND: Welcome to another episode of the Augmented Podcast. Augmented brings industrial conversations that matter, serving up the most relevant conversations on industrial tech. Our vision is a world where technology will restore the agility of frontline workers.

In this episode of the podcast, the topic is Industrial AI. Our guest is Professor Jay Lee, the Ohio Eminent Scholar, and the L.W. Scott Alter Chair Professor in Advanced Manufacturing, and the Founding Director of the Industrial AI Center at the University of Cincinnati.

In this conversation, we talk about how AI does many things but to be applicable, industry needs it to work every time, which puts on additional constraints on what can be done by when.

Augmented is a podcast for industrial leaders, process engineers, and shop floor operators hosted by futurist Trond Arne Undheim and presented by Tulip.

Jay, it's a pleasure to have you here. How are you today?

JAY: Good. Thank you for inviting me to have a good discussion about industrial AI.

TROND: Yeah, I think it will be a good discussion. Look, Jay, you are such an accomplished person, both in terms of your academics and your industrial credentials. I wanted to quickly just go through where you got to where you are because I think, especially in your case, it's really relevant to the kinds of findings and the kinds of exploration that you're now doing.

You started out as an engineer. You have a dual degree. You have a master's in industrial management also. And then you had a career in industry, worked at real factories, GM factories, Otis elevators, and even on Sikorsky helicopters. You had that background, and then you went on to do a bunch of different NSF grants. You got yourself; I don't know, probably before that time, a Ph.D. in mechanical engineering from Columbia.

The rest of your career, and you correct me, but you've been doing this mix of really serious industrial work combined with academics. And you've gone a little bit back and forth. Tell me a little bit about what went into your mind as you were entering the manufacturing topics and you started working in factories. Why have you oscillated so much between industry and practice? And tell me really this journey; give me a little bit of specifics on what brought you on this journey and where you are today.

JAY: Well, thank you for talking about this career because I cut my teeth from the factory early years. And so, I learned a lot of fundamental things in early years of automation. In the early 1980s, in the U.S, it was a tough time trying to compete with the Japanese automotive industry. So, of course, the Big Three in Detroit certainly took a big giant step, tried to implement a very good manufacturing automation system.

So I was working for Robotics Vision System at that time in New York, in Hauppage, New York, Long Island. And shortly, later on, it was invested by General Motors. And in the meantime, I was studying part-time in Columbia for my mechanical engineering, Doctor of Engineering. And, of course, later on, I transferred to George Washington because I had to make a career move. So I finished my Ph.D. Doctor of Science in George Washington later.

But the reason we stopped working on that is because of the shortage of knowledge in making automation work in the factory. So I was working full-time trying to implement the robots automation in a factory. In the meantime, I also found a lack of knowledge on how to make a robot work and not just how to make a robot move. Making it move means you can program; you can do very fancy motion. But that's not what factories want.

What factories really want is a non-stop working system so they can help people to accomplish the job. So the safety, and the certainty, the accuracy, precision, maintenance, all those things combined together become a headache actually. You have to calibrate the robot all the time. You have to reprogram them.

So eventually, I was teaching part-time in Stony Brook also later on how to do the robotic stuff. And I think that was the early part of my career. And most of the time I spent in factory and still in between the part-time study and part-time working.

But later on, I got a chance to move to Washington, D.C. I was working for U.S. Postal Service headquarters as Program Director for automation. In 1988, post service started a big initiative trying to automate a 500 mil facility in the U.S. There are about 115 number one facilities which is like New York handled 8 million mail pieces per day at that time; you're talking about '88. But most are manual process, so packages.

So we started developing the AI pattern recognition, hand-written zip code recognition, robotic postal handling, and things like that. So that was the opportunity that attracted me actually to move away from automotive to service industry. So it was interesting because you are working with top scientists from different universities, different companies to make that work. So that was the early stage of the work.

Later on, of course, I had a chance to work with the National Science Foundation doing content administration in 1991. That gave me the opportunity to work with professors in universities, of course. So then, by working with them, I was working on a lot of centers like engineering research centers and also the Industry-University Cooperative Research Centers Program, and later on, the materials processing manufacturing programs.

So 1990 was a big time for manufacturing in the United States. A lot of government money funded the manufacturer research, of course. And so we see great opportunity, like, for example, over the years, all the rapid prototyping started in 1990s. It took about 15-20 years before additive manufacturing came about. So NSF always looks 20 years ahead, which is a great culture, great intellectual driver. And also, they're open to the public in terms of the knowledge sharing and the talent and the education.

So I think NSF has a good position to provide STEM education also to allow academics, professors to work with industry as well, not just purely academic work. So we support both sides. So that work actually allowed me to understand what is real status in research, in academics, also how far from real implementation.

So in '95, I had the opportunity to work in Japan actually. I had an opportunity...NSF had a collaboration program with the MITI government in Japan. So I took the STA fellowship called science and technology fellow, STA, and to work in Japan for six months and to work with 55 organizations like Toyota, Komatsu, Nissan, FANUC, et cetera.

So by working with them, then you also understand what the real technology level Japan was, Japanese companies were. So then you got calibration in terms of how much U.S. manufacturing? How much Japanese manufacturing? So that was in my head, actually. I had good weighting factors to see; hmm, what's going on here between these two countries? That was the time.

So when I came back, I said, oh, there's something we have to do differently. So I started to get involved in a lot of other things. In 1998, I had the opportunity to work for United Technologies because UTC came to see me and said, "Jay, you should really apply what you know to real companies." So they brought me to work as a Director for Product Environment Manufacturing Department for UTRC, United Technology Research Center, in East Hartford. Obviously, UTC business included Pratt & Whitney jet engines, Sikorsky helicopters, Otis elevators, Carrier Air Conditioning systems, Hamilton Sundstrand, et cetera.

So all the products they're worldwide, but the problem is you want to support global operations. You really need not just the knowledge, what you know, but also the physical usage, what you don't know. So you know, and you don't know. So how much you don't know about a product usage, that's how the data is supposed to be coming back. Unfortunately, back in 1999, I have to tell you; unfortunately, most of the product data never came back. By the time it got back, it is more like a repair overhaul recur every year to a year later. So that's not good.

So in Japan, I was experimenting the first remote machine monitoring system using the internet actually in 1995. So I published a paper in '98 about how to remotely use physical machine and cyber machine together. In fact, I want to say that's the first digital twin but as a cyber-physical model together. That was in my paper in 1998 in Journal of Machine Tools and Manufacture.

TROND: So, in fact, you were a precursor in so many of these fields. And it just strikes me that as you're going through your career here, there are certain pieces that you seem to have learned all along the way because when you are a career changer oscillating between public, private, semi-private, research, business, you obviously run the risk of being a dilettante in every field, but you seem to have picked up just enough to get on top of the next job with some insight that others didn't have. And then, when you feel like you're frustrated in that current role, you jump back or somewhere else to learn something new.

It's fascinating to me because, obviously, your story is longer than this. You have startup companies with your students and others in this business and then, of course, now with the World Economic Forum Lighthouse factories and the work you've been doing for Foxconn as well. So I'm just curious.

And then obviously, we'll get to industrial AI, which is so interesting in your perspective here because it's not just the technology of it; it is the industrial practice of this new domain that you have this very unique, practical experience of how a new technology needs to work. Well, you tell me, how did you get to industrial AI? Because you got there to, you know, over the last 15-20 years, you integrated all of this in a new academic perspective.

JAY: Well, that's where we start. So like I said earlier, I realized industry we did not have data back in the late 1990s. And in 1999, dotcom collapsed, remember?

TROND: Yes, yes.

JAY: Yeah. So all the companies tried to say, "Well, we're e-business, e-business, e-commerce, e-commerce," then in 2000, it collapsed. But the reality is that people were talking about e-business, but in the real world, in industrial setting, there's no data almost. So I was thinking, I mean, it's time I need to think about how to look at data-centric perspectives, how to develop such a platform, and also analytics to support if one-day data comes with a worry-free kind of environment. So that's why I decided to transition to an academic career in the year 2000.

So what I started thinking, in the beginning, was where has the most data? As we all know, the product lifecycle usage is out there. You have lots of data, but we're not collecting it. So eventually, I called a central Intelligent Maintenance System called IMS, not intelligent manufacturing system because maintenance has lots of usage data which most developers of a product don't know. But if we have a way to collect this data to analyze and predict, then we can guarantee the product uptime or the value creation, and then the customer will gain most of the value back.

Now we can use the data feedback to close-loop design. That was the original thinking back in the year 2000, which at that time, no cell phone could connect to the internet. Of course, nobody believed you. So we used a term called near-zero downtime, near-zero downtime, ZDT. Nobody believed us. Intel was my first founding member. So I made a pitch to FANUC in 2001. Of course, they did not believe it either. Of course, FANUC in 2014 adopted ZDT, [laughs] ZDT as a product name.

But as a joke, when I talked to the chairman, the CEO of the company in 2018 in Japan, Inaba-san that "Do you know first we present this ZDT to your company in Michigan? They didn't believe it. Now you guys adopted." "Oh, I didn't know you use it." So when he came to visit in 2019, they brought the gift. [laughs]

So anyway, so what happened is during the year, so we worked with the study of 6 companies, 20 companies and eventually they became over 100 companies. And in 2005, I worked with Procter & Gamble and GE Aircraft Engine. They now became GE Aviation; then, they got a different environment.

So machine learning became a typical thing you use every day, every program, but we don't really emphasize AI at that time. The reason is machine learning is just a tool. It's an algorithm like a support-vector machine, self-organizing map, and logistic regression. All those are just supervised learning or now supervised learning techniques. And people use it. We use it like standard work every day, but we don't talk about AI.

But over the years, when you work with so many companies, then you realize the biggest turning point was Toyota 2005 and P&G in 2006. The reason I'm telling you 2005 is Toyota had big problems in the factory in Georgetown, Kentucky, where the Camry factory is located. So they had big compressor problems. So we implemented using machine learning, the support-vector machine, and also principal component analysis. And we enable that the surge of a compressor predicted and avoided and never happened. So until today --

TROND: So they have achieved zero downtime after that project, essentially.

JAY: Yeah. So that really is the turning point. Of course, at P&G, the diaper line continues moving the high volume. They can predict things, reduce downtime to 1%. There's a lot of money. Diaper business that is like $10 billion per year.

TROND: It's so interesting you focus on downtime, Jay, because obviously, in this hype, which we'll get to as well, people seem to focus so much on fully automated versus what you're saying, which is it doesn't really, you know, we will get to the automation part, but it is the downtime that's where a lot of the savings is obviously. Because whether it's a lights out or lights on, humans are not the real saving here. And the real accomplishment is in zero downtime because that is the industrialization factor. And that is what allows the system to keep operating. Of course, it has to do with automation, but it's not just that.

Can you then walk us through what then became industrial AI for you? Because as I've now understood it, it is a highly specific term to you. It's not just some sort of fluffy idea of very, very advanced algorithms and robots running crazy around autonomously. You have very, very specific system elements. And they kind of have to work together in some architectural way before you're willing to call it an industrial AI because it may be a machine tool here, and a machine tool there, and some data here.

But for you, unless it's put in place in a working architecture, you're not willing to call it, I mean, it may be an AI, but it is not an industrial AI. So how did this thinking then evolve for you? And what are the elements that you think are crucial for something that you even can start to call an industrial AI? Which you now have a book on, so you're the authority on the subject.

JAY: Well, I think the real motivation was after you apply all the machine learning toolkits so long...and a company like National Instruments, NI, in Austin, Texas, they licensed our machine learning toolkits in 2015. And eventually, in 2017, they started using the embedding into LabVIEW version. So we started realizing, actually, the toolkit is very important, not just from the laboratory point of view but also from the production and practitioners' point of view from industry. Of course, researchers use it all the time for homework; I mean, that’s fine.

So eventually, I said...the question came to me about 2016 in one of our industry advisory board meeting. You have so many successes, but the successes that happen can you repeat? Can you repeat? Can you repeatably have the same success in many, many other sites? Repeatable, scalable, sustainable, that's the key three keywords. You cannot just have a one-time success and then just congratulate yourself and forget it, no. So eventually, we said, oh, to make that repeat sustainable, repeatable, you have a systematic discipline.

TROND: I'm so glad you say this because I have taken part in a bunch of best practice schemes and sometimes very optimistically by either an industry association or even a government entity. And they say, "Oh yeah, let's just all go on a bunch of factory visits." Or if it's just an IT system, "Let's just all write down what we did, and then share it with other people." But in fact, it doesn't seem to me like it is that easy.

It's not like if I just explain what I think I have learned; that's not something others can learn from. Can you explain to me what it really takes to make something replicable? Because you have done that or helped Foxconn do that, for example. And now you're obviously writing up case studies that are now shared in the World Economic Forum across companies.

But there's something really granular but also something very systemic and structured about the way things have to be explained in order to actually make it repeatable. What is the sustainability factor that actually is possible to not just blue copy but turn it into something in your own factory?

JAY: Well, I think that there are basically several things. The data is one thing. We call it the data technology, DT, and which means data quality evaluation. How do you understand what to use, what not to use? How do you know which data is useful? And how do you know where the data is usable?

It doesn't mean useful data is usable, just like you have a blood donation donor, but the blood may not be usable if the donor has HIV. I like to use an analogy like food. You got a fish in your hand; wow, great. But you have to ask where the fish comes from. [chuckles] If it comes from polluted water, it's not edible, right? So great fish but not edible.

TROND: So there's a data layer which has to be usable, and it has to be put somewhere and put to use. It actually then has to be used. It can't just be theoretically usable.

JAY: So we have a lot of useful data people collect. The problem is people never realized lots of them are not usable because of a lack of a label. They have no background, and they're not normalized. So eventually, that is a problem. And even if you have a lot of data, it doesn't mean it is usable.

TROND: So then I guess that's how you get to your second layer, which I guess most people just call machine learning, but for you, it's an algorithmic layer, which is where some of the structuring gets done and some of the machines that put an analysis on this, put in place automatic procedures.

JAY: And machine learning to me it's like cooking ware like a kitchen. You got a pan fry; you got a steamer; you got the grill. Those are tools to cook the food, the data. Food is like data. Cooking ware is like AI. But it depends on purpose. For example, you want fish. What do you want to eat first? I want soup. There's a difference. Do you want to grill? Do you want to just deep fry? So depending on how you want to eat it, the cooking ware will be selected differently.

TROND: Well, and that's super interesting because it's so easy to say, well, all these algorithms and stuff they're out there, and all you have to do is pick up some algorithms. But you're saying, especially in a factory, you can't just pick any tool. You have to really know what the effect would be if you start to...for example, on downtime, right?

Because I'm imagining there are very many advanced techniques that could be super advanced, but they are perhaps not the right tool for the job, for the workers that are there. So how does that come into play? Are these sequential steps, by the way? So once you figure out what the data is then, you start to fiddle with your tools.

JAY: Well, there are two perspectives; one perspective is predict and prevent. So you predict something is going to happen. You prevent it from happening, number one. Number two, understand the root causes and potential root causes. So that comes down to the visible and invisible perspective.

So from the visible world, we know what to measure. For example, if you have high blood pressure, you measure blood pressure every day, but that may not be the reason for high blood pressure. It may be because of your DNA, maybe because of the food you eat, because of lack of exercise, because of many other things, right?

TROND: Right.

JAY: So if you keep measuring your blood pressure doesn't mean you have no heart attack. Okay, so if you don't understand the reason, measuring blood pressure is not a problem. So I'm saying that you know what you don't know. So we need to find out what you don't know. So the correlation of invisible, I call, visible-invisible. So I will predict, but you also want to know the invisible reason relationship so you can prevent that relationship from happening. So that is really called deep mining those invisibles.

So we position ourselves very clearly between visible-invisible. A lot of people just say, "Oh, we know what the problem is." The problem is not a purpose. For example, the factory manufacturing there are several very strong purposes, number one quality, right? Worry-free quality.

Number two, your efficiency, how much you produce per dollar. If you say that you have great quality, but I spent $10,000 to make it, it is very expensive. But if you spend $2 to make it, wow, that's great. How did you do it? So quality per dollar is a very different way of judging how good you are. You got A; I spent five days studying. I got A; I spent two hours studying. Now you show the capability difference.

TROND: I agree. And then the third factor in your framework seems to be platform. And that's when I think a lot of companies go wrong as well because platform is...at least historically in manufacturing, you pick someone else's platform. You say I'm going to implement something. What's available on the market, and what can I afford, obviously? Or ideally, what's the state of the art? And I'll just do that because everyone seems to be doing that. What does platform mean to you, and what goes into this choice? If you're going to create this platform for industrial AI, what kind of a decision is that?

JAY: So DT is data, AT is algorithm, and PT is platform, PT platform. Platform means some common things are used in a shared community. For example, kitchen is a platform. You can cook. I can cook. I can cook Chinese food. I can cook Italian food. I can cook Indian food. Same kitchen but different recipe, different seasoning, but same cooking ware.

TROND: Correct. Well, because you have a good kitchen, right?

JAY: Yes.

TROND: So that's --

JAY: [laughs]

TROND: Right?

JAY: On the platform, you have the most frequently used tool, not everything. You don't need 100 cooking ware in your kitchen. You probably have ten or even five most daily used.

TROND: Regardless of how many different cuisines you try to cook.

JAY: Exactly. That's called the AI machine toolkit. So we often work with companies and say, "You don't need a lot of tools, come on. You don't need deep learning. You need a good logistic regression and support-vector machine, and you're done."

TROND: Got it.

JAY: Yeah, you don't need a big chainsaw to cut small bushes. You don't need it.

TROND: Right. And that's a very different perspective from the IT world, where many times you want the biggest tool possible because you want to churn a lot of data fast, and you don't really know what you're looking for sometimes. So I guess the industrial context here really constrains you. It's a constraint-based environment.

JAY: Yes. So industry, like I said, the industry we talked about three Ps like I said: problems, purposes, and processes. So normally, problem comes from...the main thing is logistic problems, machine, and factory problems, workforce problems, the quality problems, energy problem, ignition problem, safety problems. So the problem happens every day. That's why in factory world, we call it firefighting. Typically, you firefight every day.

TROND: And is that your metaphor for the last part of your framework, which is actually operation? So operation sounds really nice and structured, right?

JAY: [chuckles] Yes.

TROND: As if that was like, yeah, that's the real thing, process. We got this. But in reality, it feels sometimes, to many who are operating a factory; it's a firefight.

JAY: Sometimes the reason lean theme work, Six Sigma, you turn a problem into a process, five Ss process, okay? And fishbone diagram, Pareto chart, and Kaizen before and after. So all the process, SOP, so doesn't matter which year workforce comes in, they just repeat, repeat, repeat, repeat, repeat.

So in Toyota, the term used to be called manufacturing is just about the discipline. It's what they said. The Japanese industry manufacturing is about discipline, how you follow a discipline to everyday standard way, sustainable way, consistent way, and then you make good products. This is how the old Toyota was talking about, old one. But today, they don't talk that anymore. Training discipline is only one thing; you need to understand the value of customers.

TROND: Right. So there are some new things that have to be added to the lean practices, right?

JAY: Yes.

TROND: As time goes by. So talk to me then more about the digital element because industrial AI to you, clearly, there's a very clear digital element, but there's so many, many other things there. So I'm trying to summarize your framework. You have these four factors: data, algorithms, platforms, and operations. These four aspects of a system that is the challenge you are dealing with in any factory environment.

And some of them have to do with digital these days, and others, I guess, really have to do more with people. So when that all comes together, do you have some examples? I don't know, we talked about Toyota, but I know you've worked with Foxconn and Komatsu or Siemens. Can you give me an example of how this framework of yours now becomes applied in a context? Where do people pick up these different elements, and how do they use them?

JAY: There's a matrix thinking. So horizontal thinking is a common thing; you need to have good digital thread including DT, data technology, AT, algorithms or analytics, PT, platform, edge cloud, and the things, and OT operation like scheduling, optimizations, stuff like that.

Now, you got verticals, quality vertical, cost vertical, efficiency verticals, safety verticals, emission verticals. So you cannot just talk about general. You got to have focus on verticals. For example, let me give you one example: quality verticals. Quality is I'm the factory manager. I care about quality. Yes, the customer will even care more, so they care. But you have a customer come to your shop once a month to check. You ask them, "Why you come?" "Oh, I need to see how good your production." "How about you don't have to come? You can see my entire quality." "Wow, how do I do that?"

So eventually, we develop a stream of quality code, SOQ, Stream Of Quality. So it's not just about the product is good. I can go back to connect all the processes of the quality segment of each station. Connect them together. Just like you got a fish, oh, okay, the fish is great. But I wonder, when the fish came out of water, when the fish was in the truck, how long was it on the road? And how long was it before reaching my physical distribution center and to my home?

So if I have a sensor, I can tell you all the temperature history inside the box. So when you get your fish, you take a look; oh, from the moment the fish came out of the boat until it reached my home, the temperature remained almost constant. Wow. Now you are worry-free. It's just one thing. So you connect together. So that's why we call SOQ, Stream Of Quality, like a river connected.

So by the time a customer gets a quality product, they can trace back and say, "Wow, good. How about if I let you see it before you come? How about you don't come?" I say, "Oh, you know what? I like it." That's what this type of manufacturing is about. It just doesn't make you happy. You have to make the customer happy, worry-free.

MID-ROLL AD:

Here's what Klaus Schwab, Executive Chairman of the World Economic Forum, says about the book: "Augmented Lean is an important puzzle piece in the fourth industrial revolution."

Find out more on www.augmentedlean.com and pick up the book in a bookstore near you.

TROND: So, Jay, you took the words out of my mouth because I wanted to talk about the future. I'm imagining when you say worry-free, I mean, you're talking about a soon-to-be state of manufacturing. Or are you literally saying there are some factories, some of the excellence factories where you've won awards in the World Economic Forum or other places that are working towards this worry-free manufacturing, and to some extent, they have achieved it?

Well, elaborate for me a little bit about the future outlook of manufacturing and especially this people issue because you know that I'm engaged...The podcast is called Augmented Podcast. I'm engaged in this debate about automation. Well, is there a discrepancy between automation and augmentation? And to what extent is this about people running the system? Or is it the machines that we should optimize to run all the system?

For you, it's all about worry-free. First of all, just answer this question, is worry-free a future ideal, or is it actually here today if you just do the right things?

JAY: Well, first of all, worry-free is our mindset where the level of satisfaction should be, right?

TROND: Yep.

JAY: So to make manufacturing happen is not about how to make good quality, how to make people physically have less worry, how to make customers less worry is what is. But the reason we have a problem with workforce today, I mean, we have a hard time to hire not just highly skilled workers but even regular workforce.

Because for some reason, not just U.S., it seems everywhere right now has similar problems. People have more options these days to select other living means. They could be an Uber driver. [laughs] They could be...I don't know. So there are many options. You don't have to just go to the factory to make earnings. They can have a car and drive around Uber and Lyft or whatever. They can deliver the food and whatever. So they can do many other things.

And so today, you want to make workforce work environment more attractive. You have to make sure that they understand, oh, this is something they can learn; they can grow. They are fulfilled because the environment gives them a lot of empowerment. The vibe, the environment gives them a wow, especially young people; when you attract them from college, they'd like a wow kind of environment, not just ooh, okay. [laughs]

TROND: Yeah. Well, it's interesting you're saying this. I mean, we actually have a lack of workers. So it's not just we want to make factories full of machines; it's actually the machines are actually needed just because there are no workers to fill these jobs. But you're looking into a future where you do think that manufacturing is and will be an attractive place going forward. That seems to be that you have a positive vision of the future we're going into. You think this is attractive. It's interesting for workers.

JAY: Yeah. See, I often say that there are some common horizontal we have to use all the day. Vertical is the purpose, quality. I talked about vertical quality first, quality. But what are the horizontal common? I go A, B, C, D, E, F. What's A? AI. B is big data. C is cyber and cloud. D is digital or digital twin, whatever. E is environment ecosystem and emission reduction. What's F? Very important, fun. [laughs] If you miss that piece, who wants to work for a place there's no fun?

You tell me would you work for...you and I, we're talking now because it's fun. You talk to people and different perspectives. I talk to you, and I say, wow, you've built some humongous network here in the physical...the future of digital, not just professional space but also social space but also the physical space. So, again, the fun things inspire people, right?

TROND: They do. So talking about inspiring people then, Jay, if you were to paint a picture of this future, I guess, we have talked just now about workers and how if you do it right, it's going to be really attractive workplaces in manufacturing. How about for, I guess, one type of worker, these knowledge workers more generally? Or, in fact, is there a possibility that you see that not just is it going to be a fun place to be for great, many workers, but it's actually going to be an exciting knowledge workplace again?

Which arguably, industrialization has gone through many stages. And being in a factory wasn't always all that rosy, but it was certainly financially rewarding for many. And it has had an enormous career progression for others who are able to find ways to exploit this system to their benefit. How do you see that going forward?

Is there a scope, is there a world in which factory work can or perhaps in an even new way become truly knowledge work where all of these industrial AI factors, the A to the Fs, produce fun, but they produce lasting progression, and career satisfaction, empowerment, all these buzzwords that everybody in the workplace wants and perhaps deserves?

JAY: That's how we look at the future workforce is not just about the work but also the knowledge force. So basically, the difference is that people come in, and they become seasoned engineers, experienced engineers. And they retire, and the wisdom carries with them. Sometimes you have documentation, Excel sheet, PPT in the server, but nobody even looks at it. That's what today's worry is.

So now what you want is living knowledge, living intelligence. The ownership is very important. For example, I'm a worker. I develop AI, not just the computer software to help the machine but also help me. I can augment the intelligence. I will augment it. When I make the product happen, the inspection station they check and just tell me pass or no pass. They also tell me the quality, 98, 97, but you pass. And then you get your score. You got a 70, 80, 90, but you got an A. 99, you got an A, 91, you got an A, 92. So what exactly does A mean?

So, therefore, I give you a reason, oh, this is something. Then I learn. Okay, I can contribute. I can use voice. I can use my opinion to augment that no, labeled. So next time people work, oh, I got 97. And so the reason is the features need to be maintained, to be changed, and the system needs to be whatever. So eventually, you have a human contribute.

The whole process could be consisting of 5 experts, 7, 10, 20, eventually owned by 20 people. That legacy continues. And you, as a worker, you feel like you're part of the team, leave a legacy for the next generation. So eventually, it's augmented intelligence.

The third level will be actual implementation. So AI is not about artificial intelligence; it is about actual implementation. So people physically can implement things in a way they can make data to decisions. So their decision mean I want to make an adjustment. I want to find out how much I should adjust. Physically, I can see the gap. I can input the adjustment level.

The system will tell me physically how could I improve 5%. Wow, that's good. I made a 5% improvement. Your boss also knows. And your paycheck got the $150 increase this month. Why? Because my contribution to the process quality improved, so I got the bonus. That's real-world feedback.

TROND: Let me ask you one last question about how this is going to play out; I mean, in terms of how the skilling of workers is going to allow this kind of process. A lot of people are telling me about the ambitions that I'm describing...and some of the guests on the podcasts and also the Tulip software platform, the owner of this podcast, that it is sometimes optimistic to think that a lot of the training can just be embedded in the work process. That is obviously an ideal.

But in America, for example, there is this idea that, well, you are either a trained worker or an educated worker, or you are an uneducated worker. And then yes, you can learn some things on the job. But there are limits to how much you can learn directly on the job. You have to be pulled out, and you have to do training and get competencies.

As you're looking into the future, are there these two tracks? So you either get yourself a short or long college degree, and then you move in, and then you move faster. Or you are in the factory, and then if you then start to want to learn things, you have to pull yourself out and take courses, courses, courses and then go in? Or is it possible through these AI-enabled training systems to get so much real-time feedback that a reasonably intelligent person actually never has to be pulled out of work and actually they can learn on the job truly advanced things?

So because there are two really, really different futures here, one, you have to scale up an educational system. And, two, you have to scale up more of a real-time learning system. And it seems to me that they're actually discrepant paths.

JAY: Sure. To me, I have a framework in my book. I call it the four P structure, four P. First P is principle-based. For example, in Six Sigma, in lean manufacturing, there's some basic stuff you have to study, basic stuff like very simple fishbone diagram. You have to understand those things. You can learn by yourself what that is. You can take a very basic introduction course. So we can learn and give you a module. You can learn yourself or by a group, principle-based.

The second thing is practice-based. Basically, we will prepare data for you. We will teach you how to use a tool, and you will do it together as a team or as individual, and you present results by using data I give to you, the tool I give to you. And it's all, yeah, my team A presented. Oh, they look interesting. And group B presented, so we are learning from each other.

Then after the group learning is finished, you go back to your team in the real world. You create a project called project-based learning. You take a tool you learn. You take the knowledge you learn and to find a project like a Six Sigma project you do by yourself. You formulate. And then you come back to the class maybe a few weeks later, present with a real-world project based on the boss' approval.

So after that, you've got maybe a black belt but with the last piece professional. Then you start teaching other people to repeat the first 3ps. You become master black belt. So we're not reinventing a new term. It really is about a similar concept like lean but more digital space. Lean is about personal experience, and digital is about the data experience is what's the big difference.

TROND: But either way, it is a big difference whether you have to rely on technological experts, or you can do a lot of these things through training and can get to a level of aptitude that you can read the signals at least from the system and implement small changes, perhaps not the big changes but you can at least read the system.

And whether they're low-code or no-code, you can at least then through learning frameworks, you can advance, and you can improve in not just your own work day, but you can probably in groups, and feedbacks, and stuff you can bring the whole team and the factory forward perhaps without relying only on these external types of expertise that are actually so costly because they take you away. So per definition, you run into this; I mean, certainly isn't worry-free because there is an interruption in the process.

Well, look, this is fascinating. Any last thoughts? It seems to me that there are so many more ways we can dig deeper on your experience in any of these industrial contexts or even going deeper in each of the frameworks. Is there a short way to encapsulate industrial AI that you can leave us with just so people can really understand?

JAY: Sure.

TROND: It's such a fundamental thing, AI, and people have different ideas about that, and industry people have something in their head. And now you have combined them in a unique way. Just give us one sentence: what is industrial AI? What should people leave this podcast with?

JAY: AI is a cognitive science, but industrial AI is a systematic discipline is one sentence. So that means people have domain knowledge. Now we have to create data to represent our domain then have the discipline to solve the domain problems. Usually, with domain knowledge, we try with our experience, and you and I know; that's it. But we have no data coming out. But if I have domain become data and data become discipline, then other people can repeat our success even our mistake; they understand why. So eventually, domain, data, discipline, 3 Ds together, you can make a good decision, sustainable and long-lasting.

TROND: Jay, this has been so instructive. I thank you for spending this time with me. And it's a little bit of a never-ending process.

JAY: [laughs]

TROND: Industry is not something that you can learn it and then...because also the domain changes and what you're doing and what you're producing changes as well. So it's a lifelong --

JAY: It's rewarding.

TROND: Rewarding but lifelong quest.

JAY: Yeah. Well, thank you for the opportunity to share, to discuss. Thank you.

TROND: It's a great pleasure.

You have just listened to another episode of the Augmented Podcast with host Trond Arne Undheim. The topic was Industrial AI. And our guest was Professor Jay Lee from University of Cincinnati. In this conversation, we talked about how AI in industry needs to work every time and what that means.

My takeaway is that industrial AI is a breakthrough that will take a while to mature. It implies discipline, not just algorithms. In fact, it entails a systems architecture consisting of data, algorithm, platform, and operation.

Thanks for listening. If you liked the show, subscribe at augmentedpodcast.co or in your preferred podcast player, and rate us with five stars.

If you liked this episode, you might also like Episode 81: From Predictive to Diagnostic Manufacturing Augmentation. Hopefully, you'll find something awesome in these or in other episodes, and if so, do let us know by messaging us. We would love to share your thoughts with other listeners.

The Augmented Podcast is created in association with Tulip, the frontline operation platform that connects the people, machines, devices, and systems used in a production or logistics process in a physical location. Tulip is democratizing technology and is empowering those closest to operations to solve problems. Tulip is also hiring. You can find Tulip at tulip.co.

Please share this show with colleagues who care about where industry and especially where industrial tech is heading. To find us on social media is easy; we are Augmented Pod on LinkedIn and Twitter and Augmented Podcast on Facebook and YouTube.

Augmented — industrial conversations that matter. See you next time. Special Guest: Jay Lee.