The Derby Mill Series

Nvidia CEO Jensen Huang recently described physical AI, a category that includes robots that can perceive, understand and act in the real world, as the next wave in artificial intelligence. So in the last episode before Derby Mill’s summer break, and our first-ever in-person recording session, the team of Ajay Agrawal, Rich Sutton, Sendhil Mullainathan and Niamh Gavin welcome Suzanne Gildert, the CEO and founder of Nirvanic Consciousness Technologies in Vancouver, BC.Gildert is a pioneering figure in the humanoid robotics community. Here, she discusses with the Derby Mill team such questions as: Why now for humanoid robots? What are the advantages and disadvantages of the bipedal human form factor? What makes humanoid robots difficult to create? The episode concludes with Ajay asking the team what they want listeners to think about during our two-month summer break. See you in September!GUESTS AND HOSTSSuzanne Gildert, co-founder & CEO, Nirvanic Consciousness TechnologiesAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist, CEO, Emergent PlatformsLINKSNirvanic Consciousness Technologies homepageThe Jenson Huang / NVIDIA presentation Ajay references early in the episode. Reuters storyBoth Sendhil and Rich love the sci-fi novels of Iain BanksDerby Mill show websiteRich Sutton’s home page. Follow Rich on XSendhil Mullainathan’s website. Follow Sendhil on XBe sure to catch every episode of The Derby Mill Series by subscribing on the following platforms:YouTube // Spotify // Apple Podcasts // SubstackDISCUSSION POINTS00:00 Cold open01:52 Welcome and intro to humanoid robots and Suzanne Gildert04:20 What’s so hard about building a humanoid robot?06:15 The complexity of the human body07:55 So why bother making a humanoid robot?09:30 Why are humanoid robots so hot right now?10:50 Why now: AI software15:49 Rich Sutton explains why humanoid robots are so intriguing17:05 Can we code robots the same way we approached LLMs?20:30 Teaching robots with reinforcement learning in simulation21:47 Sendhil: How important are humanoid robots?29:10 Niamh: Is the bipedal form factor the best all-around solution?37:15 Sendhil: What about hybrid human-robot creatures?41:15 Agent architecture and humanoid robots44:35 The idea that we explore by random action selection48:00 Suzanne on types of decision making52:13 Decision making as centrepiece of economics, and AI57:19 Quantum physics and self-aware AI1:00:50 Defining consciousness1:05:00 Lightning round: Niamh on cost of experimentation1:06:33 LR: Ajay on what’s RLable1:07:37 LR: Rich on AI disillusionment1:08:40 LR: Suzanne on AI consciousness1:10:20 LR: Sendhil on the “what is AI” turf war1:18:18 Ajay wraps up season 1NUGGETSNugget 1 - Cost of ExperimentationIntrepid's Ajay Agrawal asks AI scientist Niamh Gavin to name one topic for listeners to reflect on over Derby Mill's summer break.Nugget 2 - AI Disillusionment and Turf WarsIntrepid's Ajay Agrawal asks Turing Award winner Rich Sutton to name one topic for listeners to reflect on over Derby Mill's summer break.Nugget 3 - Consciousness and EmpathyIntrepid's Ajay Agrawal asks Nirvanic CEO Suzanne Gildert to name one topic for listeners to reflect on over Derby Mill's summer break. Her response is to appeal to viewers to question any fearful reaction they have to the notion of conscious AI.DISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Headquartered in Toronto, Private AI detects and removes personally identifiable information (PII) from data using large language models (LLMs), all without compromising individual or institutional privacy. In this episode, Private AI co-founder and CEO Patricia Thaine offers a behind-the-scenes look at the company’s technical strategy, including the scalability challenge inherent in protecting confidential company information, and the growing threat of re-identification. With more than 30,000 hours invested in building their PII detection system, Private AI now operates in seven countries and partners with organizations such as the Business Development Bank of Canada, MaRS, and the University of Toronto.This episode features the Intrepid team exploring such questions as:* How can organizations effectively protect personally identifiable information (PII) and confidential company information in large language models?* What are the risks of re-identification, even after attempting to anonymize data?* How can companies balance data utility with privacy preservation?* How can privacy protection be approached as a dynamic, evolving challenge rather than a static solution?* What role can technology play in helping organizations understand and control their data privacy?GUESTS AND HOSTSPatricia Thaine, co-founder & CEO, Private AIAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist, CEO, Emergent PlatformsLINKSPrivate AI website, explainer videoPrivate AI demo, PrivateGPTRead the NYT article, A Face Is Exposed for AOL Searcher No. 4417749Pymetrics, a company that pioneered the use of AI and behavioural science to improve workforce decisions, was acquired by Harver.Rich Sutton’s home page. Follow Rich on XSendhil Mullainathan’s website. Follow Sendhil on XBe sure to catch every episode of The Derby Mill Series by subscribing on the following platforms:YouTube // Spotify // Apple Podcasts DISCUSSION POINTS00:00 Introduction01:22 Meet Patricia Thaine of Private AI01:41 About Private AI02:54 How Private AI redacts and protects data04:18 What would scalability look like for confidential company info?08:14 Deconstructing NYT’s article: A Face is Exposed for AOL Searcher No. 441774910:12 Can your digital footprint be an identifier?15:35 Solving the synthetic data problem19:15 How data minimization can help with privacy21:24 Mapping out the future of data privacy25:56 What would a reward function look like?27:12 Final commentsNUGGETSNugget 1 - The Challenge of De-Identification Concerning Data PrivacyPrivate AI CEO and co-founder Patricia Thaine describes the challenge of data privacy and de-identification.Nugget 2 - Consumer Market Demand and RegulationIntrepid’s Sendhil Mullainathan explores the challenge of creating a start-up in the “personally identifiable information” space. Nugget 3 - Different Types of CII and PIIPrivate AI’s Patricia Thaine discusses the nuances of removing personally identifiable information, as even a piece of jewellery in an X-ray can compromise anonymity. DISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Dr. Andrew Gostine, CEO of Artisight, and Tim Koby, the Chief Science Officer, share insights on AI's transformative role in healthcare. They discuss strategies for building 'smart hospitals' that leverage AI for real-time data analysis, enhancing patient safety, and preventing falls. The conversation highlights AI's potential in early sepsis prediction and its impact on hospital efficiency. They also address the ethical implications of AI in healthcare, emphasizing the importance of human oversight in navigating these advancements.

StackAdapt is an advertising platform that leverages AI to optimize digital ad campaigns across multiple channels, including display, video, native, and connected TV. With an auction system, it evaluates millions of ad opportunities each second using predictive analytics to maximize ROI and enhance audience targeting. By integrating customer data and providing privacy-conscious, scalable solutions, StackAdapt provides advertisers with data-driven insights and automated ad placement. So how can AI enhance discovery and shape awareness of digital advertising solutions that people may not yet realize they need? And what reward systems might be most effective for RL in optimising ad campaigns? The Derby Mill team talks to StackAdapt CTO and co-founder Yang Han to discuss potential answers. GUESTS AND HOSTSYang Han, CTO and co-founder, StackAdaptAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist, CEO, Emergent PlatformsLINKSDerby Mill show websiteStackAdapt’s website and explainer videoRead Rich Sutton’s latest paper Welcome to the Era of ExperienceRich Sutton’s home page. Follow Rich on XSendhil Mullainathan’s website. Follow Sendhil on XBe sure to catch every episode of The Derby Mill Series by subscribing on the following platforms:YouTube // Spotify // Apple Podcasts DISCUSSION POINTS 00:00 Introduction02:00 Welcome, Yang Han, CTO and Co-Founder of StackAdapt02:45 How advertising on StackAdapt works09:10 How StackAdapt thinks about ROI11:30 Yang on ad competition and who gets the credit.13:55 Niamh on what StackAdapt will look like at the limit.18:23 Sendhil on how we can surface decision-making in advertising.24:20 Rich on the advantages of assistance-based shopping.29:13 Becoming customer-focused with the rise of AI31:53 What executives lack when perfecting the matching problem.26:47 What’s one thing investors should pay attention to in this industry? Nugget 01 - Alternative Customer-First Business ModelNugget 02 - Educating Customers with Personalized AdsNugget 03 - Disrupting the Ad Market with Agent DiscoveryDISCLAIMERIntrepid GP is an investor in StackAdapt. The content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Explore the fascinating shift from human-generated data to autonomous agents in artificial intelligence. Discover how experiential learning may unlock superhuman capabilities, ushering in a new era of innovation. The discussion highlights the philosophical implications of AI's evolution and its potential to transform scientific discovery. Learn about the importance of simplicity in knowledge and how grounded motivations can enhance AI development. This engaging conversation reveals the promising future of AI driven by experience.

Skin Analytics is a UK company using AI to automate the diagnosis of serious skin conditions, starting with skin cancer. Its core product, DERM, is the only Class III CE mark AI medical device for autonomous dermatology in the UK’s health system. Used on more than 150,000 real-world patients, DERM achieves 99.8% negative predictive value, outperforming dermatologists. The company is expanding into general dermatology and launching in the EU and US.In the future, Skin Analytics intends to create a dermatology AI platform that is able to diagnose and treat a broader range of conditions. Based on a diverse sampling of low-cost data, the company intends its platform to transition from self-supervised to unsupervised learning, enabling ubiquitous, low-friction health monitoring.This episode features the Intrepid team exploring such questions as:* What would it take to build healthcare around AI abundance, not human bottlenecks?* How might one frame an approach to reach 99% automation in dermatological triage?* What are the tradeoffs between sensitivity, specificity, and health system efficiency?* How could reward systems (RL or pathway-based optimization) be introduced?* What’s the potential of self-supervised learning across multiple medical modalities?GUESTS AND HOSTSNeil Daly, founder and director, Skin AnalyticsJack Greenhalgh, AI director, Skin AnalyticsAjay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, senior advisor, Intrepid Growth Partners, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, senior advisor, Intrepid Growth Partners, MacArthur Genius grant recipient and professor, MITNiamh Gavin, senior advisor, Intrepid Growth Partners, Applied AI scientist, CEO, Emergent PlatformsLINKSDerby Mill show website: insights.intrepidgp.com/podcastSkin Analytics website and explainer videoRich Sutton’s home page. Follow Rich on XSendhil Mullainathan’s website. Follow Sendhil on XBe sure to catch every episode of The Derby Mill Series by subscribing on the following platforms: YouTube // Spotify // Apple Podcasts DISCUSSION POINTS00:00 Introduction01:24 Meet the team: Skin Analytics06:12 The lead-up to image recognition10:29 Patient drop-off post-referral14:03 Getting classification right18:47 Integrating into the healthcare system22:36 Cancer detection in the limit27:55 At-home cancer detection34:10 Making dermatology RL-able45:00 Using data as proxies for other diagnoses50:21 Early detection vs. overdiagnosis55:07 Higher rates of cancer detection advantages57:00 What took so long?59:07 Final remarksNugget 01 - Sensors Reveal Hidden Data in the SkinTraditionally, dermatology has been rate-limited by the human eye and optical sensors. So incorporating a variety of additional sensors to collect more diverse and comprehensive data can open the door to a new kind of pre-primary care, potentially revealing more information about internal conditions like hypertension or liver disease.Nugget 02 - The Economic Model Behind At-Home DiagnosesThere's a massive direct-to-consumer interest in skin health, which opens the door to a potential expansion of at-home skin-monitoring apps that could be used beyond only in primary care settings. But overdiagnoses risk overwhelming the healthcare system. In order to avoid case buildup, these apps require an economic model that leverages medical systems and consumer trust.Nugget 03 - Redesigning the Treatment DelayWhat prevents people from accessing treatment is not the diagnostic delay (which often involves a lengthy wait for results), but rather the delay in seeking help: People tend to wait for a reason to address an issue, which increases the risk of lowering the survival rate as a disease spreads.DISCLAIMERIntrepid GP is an investor in Skin Analytics. The content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

In this unpacked episode, the team further expands its discussion of themes that came up in episode seven, which explored the automation of customer support with artificial intelligence. Our guests in that episode were a duo that is leading efforts in that space: CEO Mike Murchison and chief product and technology officer Mike Gozzo from Ada.In this episode, Intrepid Growth Partners cofounder and partner Ajay Agrawal leads the discussion with Intrepid Senior Advisors Rich Sutton (Turing Award winner), Sendhil Mullainathan (MacArthur genius grant recipient) and Niamh Gavin (CEO, Emergent Platforms).In the previous episode, we learned that Ada’s north star is “percent automated resolutions”, or the percentage of customer inquiries that are fully resolved by AI without human intervention. One challenge is that Ada relies on large language models (LLMs) rather than action-based goals, often requiring human agents to step in when confidence is low.“It’s a mistake to think that [Ada’s AI agents] have goals,” says Sutton. ”What we have instead … is we have [AI agents] mimicking people.”All of which raises the question of how customer support will evolve as this technology advances towards the limit.Our team also debates the need for clear, objective measures of AI performance and the challenges of achieving true goal-oriented AI systems.Our panel of experts:Ajay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, MacArthur Genius grant recipient and professor, MITNiamh Gavin, Applied AI scientist, CEO, Emergent PlatformsSutton, Mullainathan and Gavin are all Intrepid Growth Partners’ senior advisors.LINKSAda websiteThis episode extends the discussion from Derby Mill episode 07: Customer Support Rich Sutton’s home page. Follow Rich on XSendhil Mullainathan’s website. Follow Sendhil on XBe sure to catch every episode by subscribing on the following platforms:YouTube // Spotify // Apple PodcastsDISCUSSION POINTS00:00 Introduction and opening credit02:00 Ada refresher03:46 Clip: Testing harness06:50 Clip discussion begins10:15 What are goal-based objectives?13:30 Is this the year of the agent?17:40 What makes agents goal-oriented19:20 Decision-making fundamentals in AI21:27 Clip: Automating system improvement over time23:32 Clip discussion begins30:13 Automating the evaluation process34:08 What could Ada look like in the limit?41:03 Closing remarksNUGGET 01: Vertical vs. Horizontal CompetitionFine-tuning used to be costly and impractical, pushing companies to open-source solutions—only to revert to OpenAI due to complexity. Now, companies like Ada build on top of model providers, offering flexibility while managing AI’s complexity. Niamh discusses the competitiveness between verticalized AI (industry-specific applications) and horizontal AI (broad sector models).NUGGET 02: The Challenge of InterpretabilityAda's evaluations rely on human judgment. The challenge here is interpretability—determining whether an outcome is truly good without direct human input. Rich Sutton offers potential solutions, including using reinforcement learning with human feedback (RLHF) as a proxy measure trained on high-quality data.NUGGET 03: Benchmarking vs. Deployment in the FieldNiamh and Sendhil discuss how, despite concerns about hallucinations in AI-powered customer service, CEOs adopt GenAI more for signaling competence than for real effectiveness.DISCLAIMERThe content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

Meet Ada, a Canadian AI agent platform automating the resolution of customer service interactions. When customers have complex requests—such as resetting passwords, checking order status, or requesting a refund—Ada uses large language models to radically reduce the amount of human effort required to fulfill the customer’s inquiry.Here, Ada CEO Mike Murchison and Chief Product & Technology Officer Mike Gozzo join the Derby Mill podcast to discuss the intersection of AI and customer support—and where the technology may go, at the limit.Our panel of experts:Ajay Agrawal, co-founder and partner, Intrepid Growth PartnersRichard Sutton, 2024 Turing Award recipient, pioneer of reinforcement learning and professor, University of AlbertaSendhil Mullainathan, MacArthur Genius grant recipient and professor, MITNiamh Gavin, Applied AI scientist, CEO, Emergent PlatformsLINKSAda websiteAda CEO Mike Murchison LinkedInAda Chief Product & Technology Officer Mike Gozzo LinkedInRich Sutton’s home page. Follow Rich on X.Sendhil Mullainathan’s website. Follow Sendhil on X.Sendhil’s article on Algorithms Need Managers, Too published in the Harvard Business Review Be sure to catch every episode by subscribing on the following platforms:YouTube // Spotify // Apple PodcastsDISCUSSION POINTS00:00 Introduction01:49 Meet Ada, the company automating customer support05:05 Customer service & books: an analogy05:41 Murchison describes automated resolution08:50 Human feedback for automated improvement 23:17 LLMs in customer service26:10 The difference between language and action26:34 Ada’s use of LLMs30:01 Murchison on how “deterministic” Ada’s actions are30:59 Improving decision quality37:06 Protecting against LLM’s unreliability 44:40 Closing remarksNUGGET 01: Human Feedback for Automated ImprovementAda describes the role of humans "coaching" their AIs. Why this is one of the first areas for "automated improvement," and how can the preference data they are collecting through the coaching process be used to "drive automated improvements throughout the entire system."NUGGET 02: Decision-Making QualityRich Sutton asks how Ada improves the quality of the system's decisions, and questions the role of humans vs. AI in terms of evaluating versus improving the quality of decisions.NUGGET 03: DistillationGiven the cost and latency virtues of smaller models, when do we anticipate applications to use large foundation models at the limit? Is the Ada case a good example of using large models to bootstrap a commercial solution en route to smaller, more specialized models?DISCLAIMER The content of this podcast is for informational and educational purposes only and should not be construed as marketing, solicitation, or an offer to buy or sell any securities or investments. The opinions expressed in this video are those of the participants and do not necessarily reflect the views of Intrepid Growth Partners or its affiliates. Any discussion of specific companies, technologies, or industries is for illustrative purposes and does not constitute investment advice. Viewers are encouraged to consult with their own financial, legal, and tax advisors before making any investment decisions. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit insights.intrepidgp.com

The conversation dives into how artificial intelligence is transforming mining exploration by utilizing lower-fidelity data like aerial imagery and dust analysis. Experts discuss the parallels between mining and healthcare, highlighting how devices like the Apple Watch can offer valuable insights despite their lower accuracy. There's a focus on enhancing predictive models by integrating diverse data sources, and how AI, when combined with human intuition, can revolutionize decision-making. The discussion emphasizes the potential of cost-effective data in both geological exploration and medical diagnostics.

Grant Sanden, CEO of GeologicAI, shares insights on revolutionizing mining through AI. He discusses how advanced core-sampling technology can significantly enhance geological predictions while navigating uncertainties. The conversation dives into the unique challenges of mining data—rich in some areas yet sparse in others—and how AI helps integrate massive datasets for better decision-making. They also highlight innovative methods like using drones and IoT sensors to improve resource exploration and operational efficiency, addressing both accuracy and sustainability in mining.