Uploaded November 2025 | Updated September 2026, 1 week ago
Prof. David Krakauer argues that intelligence isn't about knowing more—it's about doing more with less.
His core thesis flips our assumptions on their head: while emergence in complex systems is about "more is different," intelligence is fundamentally about "less is more."
Why Large Language Models Miss the Point
LLMs are "more is more", Krakauer explains. They're essentially sophisticated libraries that know everything but understand nothing. When a student looks up answers in a library during an exam, we call them a cheater, not intelligent.
Intelligence Across All Life
Perhaps most surprisingly, Krakauer claims that all living things are intelligent—even bacteria. His reasoning: intelligence is the information accumulated through evolution's selective pressures. It's why we can say an elephant is "smarter" than a worm, but we'd never say it's "more alive." The difference is accumulated capacity.
Three Dimensions of Intelligence
Krakauer proposes intelligence exists in three distinct flavours:
- Strategic Intelligence - Adaptation and survival where viruses actually beat us/
- Inferential Intelligence - Math and computation (where we built calculators because we're so bad at it)
- Representational Intelligence - Finding better ways to encode problems (the most uniquely human)
The Soma Cube Insight
Using a Soma Cube puzzle, Krakauer demonstrates something incredible: a four-year-old can solve a combinatorial problem with 15,000 logical clauses—not by computing, but by using physical representation. The physical world does the computing for us. This is intelligence: making impossibly hard problems tractable through clever representation.
This talk was recorded at DISI 2025 disi.org
David Krakauer
President and William H. Miller Professor of Complex Systems
https://www.santafe.edu/people/profile/david-krakauer
Note this version is edited down, you can watch the full version on our Patreon - patreon.com/posts/masterclass-on-142898847
Watch interview we published with David - youtube.com/watch?v=jXa8dHzgV8U
TRANSCRIPT:
app.rescript.info/public/share/Cs8QFwMw_FiOEROkNPiIf6Y-e_ahAxNMl0oFiL7_4QA
TOC:
00:00:00 - Complexity, Life & Intelligence
00:00:45 - Ptolemy vs Newton
00:01:30 - Purpose of Science
00:02:15 - Feynman on Rules vs Strategies
00:03:00 - Entropy as Unifying Rule
00:03:45 - Intelligence in Physics
00:04:15 - Complex Systems & Broken Symmetries
00:05:00 - Historical Definitions of Intelligence
00:06:15 - Herbert: Capacity to Acquire Capacity
00:06:45 - Problems with Turing Test
00:07:30 - Intelligence vs Knowledge
00:08:00 - LLMs as Libraries
00:08:30 - Humanistic Perspectives
00:09:00 - Unified Theory: Life & Intelligence
00:09:30 - Intensive vs Extensive Properties
00:10:30 - Selection Gradient Analogy
00:11:15 - Universal Intelligence (Bacteria to Elephants)
00:12:30 - Three Dimensions of Intelligence
00:13:00 - Strategic Intelligence & Viruses
00:13:30 - Inferential Intelligence & Calculators
00:14:15 - Encoding & Representation
00:14:30 - Different Paths: Life vs AI
00:15:15 - Soma Cube Problem
00:16:00 - Principle of Materiality
00:17:15 - Embodied Representation
00:17:45 - Intelligence Makes Hard Problems Easy
00:18:00 - Stupidity & Conclusion
REFS:
Papers and Academic Works
"More is Different" by Phil Anderson [https://www.tkm.kit.edu/downloads/TKM1_2011_more_is_different_PWA.pdf ]
Recent paper with Melanie and John Krakauer on emergence [arxiv.org/abs/2506.11135 ]
Formal treatment paper on intelligence/evolution duality from David - Darwinian demons, evolutionary complexity, and information maximization [https://wiki.santafe.edu/images/b/b1/CHAOEH213037110_1.pdf ]
"The Unreasonable Effectiveness of Mathematics in the Natural Sciences" by Eugene Wigner [webhomes.maths.ed.ac.uk/~v1ranick/papers/wigner.pdf ]
Prof. David Krakauer argues that intelligence isn't about knowing more—it's about doing more with less.
His core thesis flips our assumptions on their head: while emergence in complex systems is about "more is different," intelligence is fundamentally about "less is more."
Why Large Language Models Miss the Point
LLMs are "more is more", Krakauer explains. They're essentially sophisticated libraries that know everything but understand nothing. When a student looks up answers in a library during an exam, we call them a cheater, not intelligent.
Intelligence Across All Life
Perhaps most surprisingly, Krakauer claims that all living things are intelligent—even bacteria. His reasoning: intelligence is the information accumulated through evolution's selective pressures. It's why we can say an elephant is "smarter" than a worm, but we'd never say it's "more alive." The difference is accumulated capacity.
Three Dimensions of Intelligence
Krakauer proposes intelligence exists in three distinct flavours:
- Strategic Intelligence - Adaptation and survival where viruses actually beat us/
- Inferential Intelligence - Math and computation (where we built calculators because we're so bad at it)
- Representational Intelligence - Finding better ways to encode problems (the most uniquely human)
The Soma Cube Insight
Using a Soma Cube puzzle, Krakauer demonstrates something incredible: a four-year-old can solve a combinatorial problem with 15,000 logical clauses—not by computing, but by using physical representation. The physical world does the computing for us. This is intelligence: making impossibly hard problems tractable through clever representation.
This talk was recorded at DISI 2025 disi.org
David Krakauer
President and William H. Miller Professor of Complex Systems
https://www.santafe.edu/people/profile/david-krakauer
Note this version is edited down, you can watch the full version on our Patreon - patreon.com/posts/masterclass-on-142898847
Watch interview we published with David - youtube.com/watch?v=jXa8dHzgV8U
TRANSCRIPT:
app.rescript.info/public/share/Cs8QFwMw_FiOEROkNPiIf6Y-e_ahAxNMl0oFiL7_4QA
TOC:
00:00:00 - Complexity, Life & Intelligence
00:00:45 - Ptolemy vs Newton
00:01:30 - Purpose of Science
00:02:15 - Feynman on Rules vs Strategies
00:03:00 - Entropy as Unifying Rule
00:03:45 - Intelligence in Physics
00:04:15 - Complex Systems & Broken Symmetries
00:05:00 - Historical Definitions of Intelligence
00:06:15 - Herbert: Capacity to Acquire Capacity
00:06:45 - Problems with Turing Test
00:07:30 - Intelligence vs Knowledge
00:08:00 - LLMs as Libraries
00:08:30 - Humanistic Perspectives
00:09:00 - Unified Theory: Life & Intelligence
00:09:30 - Intensive vs Extensive Properties
00:10:30 - Selection Gradient Analogy
00:11:15 - Universal Intelligence (Bacteria to Elephants)
00:12:30 - Three Dimensions of Intelligence
00:13:00 - Strategic Intelligence & Viruses
00:13:30 - Inferential Intelligence & Calculators
00:14:15 - Encoding & Representation
00:14:30 - Different Paths: Life vs AI
00:15:15 - Soma Cube Problem
00:16:00 - Principle of Materiality
00:17:15 - Embodied Representation
00:17:45 - Intelligence Makes Hard Problems Easy
00:18:00 - Stupidity & Conclusion
REFS:
Papers and Academic Works
"More is Different" by Phil Anderson [https://www.tkm.kit.edu/downloads/TKM1_2011_more_is_different_PWA.pdf ]
Recent paper with Melanie and John Krakauer on emergence [arxiv.org/abs/2506.11135 ]
Formal treatment paper on intelligence/evolution duality from David - Darwinian demons, evolutionary complexity, and information maximization [https://wiki.santafe.edu/images/b/b1/CHAOEH213037110_1.pdf ]
"The Unreasonable Effectiveness of Mathematics in the Natural Sciences" by Eugene Wigner [webhomes.maths.ed.ac.uk/~v1ranick/papers/wigner.pdf ]
![He won a Nobel here for AlphaFold. Then he left. - John Jumper
This episode is sponsored by Notion. Learn more about Notions Developer Platform today at https://notion.com/mlst
Protein folding stalled biology for fifty years. A sequence of amino acids dictates a three-dimensional shape, but reading that shape meant a year and roughly $100,000 of crystallography per structure. Then AlphaFold 2 won CASP14 so decisively the organizers called the problem essentially solved.
In this documentary cut, John Jumper, who shared the 2024 Nobel Prize in Chemistry and has since left DeepMind for Anthropic, walks Tim Scarfe through what the system did and, more interestingly, what it did not. The architecture gets a proper dissection: MSAs, the Evoformer, invariant point attention, the FAPE loss, and Jumpers correction of the equivariance story, which ablations valued at roughly 2.5 of 30 GDT points rather than the whole win. He is blunt about the limits. AlphaFold predicts one experiment extraordinarily well; it is not a model of the cell, it does not capture dynamics, and on a given drug target it is wrong nine times out of ten.
From there: the AlphaFold Database of 200M+ predicted structures, AlphaFold 3 and ligands, Isomorphic Labs, and Jumpers quarrel with the bitter lesson, where finite data and human hypotheses still matter. Emmanuel Nji of BioStruct Africa closes the film on what changes when work that took years now takes months, and on training the next thousand structural biologists across Africa.
TIMESTAMPS:
00:00:00 Cold open: predicting nature with a button press
00:01:03 The protein folding bottleneck and CASP
00:04:39 The Nobel, the database, and the move to Anthropic
00:05:50 Sponsor (Notion) and framing: what AlphaFold does not claim
00:07:39 Proteins as self-assembling nanomachines
00:12:24 From structures to biology: drug discovery and Midnolin
00:17:37 The humility of AlphaFold: a narrow predictor
00:22:18 Inside the architecture: Evoformer, IPA and FAPE
00:30:20 Ruthless empiricism: ablations and 100x in data
00:35:20 Predict, control, understand
00:40:00 Against the bitter lesson; AlphaFold 3 as diffusion
00:45:07 Intelligence, representations and AGI
00:49:23 Epilogue: AlphaFold in Africa
00:52:16 Closing: the case for hybrid science models
REFERENCES:
organization:
[00:01:55] Critical Assessment of Structure Prediction (CASP)
https://predictioncenter.org/
[00:04:39] The Nobel Prize in Chemistry 2024
https://www.nobelprize.org/prizes/chemistry/2024/summary/
[00:05:18] BioStruct Africa
https://www.biostructafrica.org/
[00:18:03] Isomorphic Labs
https://www.isomorphiclabs.com/
paper:
[00:03:09] AlphaFold Protein Structure Database
https://doi.org/10.1093/nar/gkab1061
[00:17:25] Accurate structure prediction of biomolecular interactions with AlphaFold 3
https://www.nature.com/articles/s41586-024-07487-w
[00:22:18] Highly accurate protein structure prediction with AlphaFold
https://www.nature.com/articles/s41586-021-03819-2
[00:23:10] Midnolin promotes degradation of substrates independent of ubiquitination
https://doi.org/10.1126/science.adh5021
[00:27:00] Improved protein structure prediction using potentials from deep learning
https://www.nature.com/articles/s41586-019-1923-7
tool:
[00:03:09] AlphaFold Protein Structure Database (EBI)
https://alphafold.ebi.ac.uk/
[00:45:55] AlphaEvolve: a coding agent for designing advanced algorithms
https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/
other:
[00:39:40] The Bitter Lesson
http://www.incompleteideas.net/IncIdeas/BitterLesson.html
ReScript: https://app.rescript.info/share/d8cde5c221fb71e2c0f5aafe94f90dfa
Disclaimer - not sponsored, editorial with us - we filmed it at GDM, London He won a Nobel here for AlphaFold. Then he left. - John Jumper](https://i.ytimg.com/vi/e3gBwLWAerw/mqdefault.jpg)


![Type a Sentence, Get a Playable 3D World in 3 Seconds - Shlomi Fuchter & Jack Parker-Holder
This episode features Shlomi Fuchter and Jack Parker Holder from Google DeepMind, who are unveiling a new AI called Genie 3. The host, Tim Scarfe, describes it as the most mind-blowing technology he has ever seen. We were invited to their offices to conduct the interview (not sponsored).
Imagine you could create a video game world just by describing it. Thats what Genie 3 does. Its an AI world model that learns how the real world works by watching massive amounts of video. Unlike a normal video game engine (like Unreal or the one for Doom) that needs to be programmed manually, Genie generates a realistic, interactive, 3D world from a simple text prompt.
**SPONSOR MESSAGES***
Prolific: Quality data. From real people. For faster breakthroughs.
https://prolific.com/mlst?utm_campaign=98404559-MLST&utm_source=youtube&utm_medium=podcast&utm_content=script-gen
***
Here’s a breakdown of what makes it so revolutionary:
From Text to a Virtual World: You can type a drone flying by a beautiful lake or a ski slope, and Genie 3 creates that world for you in about three seconds. You can then navigate and interact with it in real-time.
Its Consistent: The worlds it creates have a reliable memory. If you look away from an object and then look back, it will still be there, just as it was. The guests explain that this consistency isnt explicitly programmed in; its a surprising, emergent capability of the powerful AI model.
A Huge Leap Forward: The previous version, Genie 2, was a major step, but it wasnt fast enough for real-time interaction and was much lower resolution.
Genie 3 is 720p, interactive, and photorealistic, running smoothly for several minutes at a time.
The Killer App - Training Robots: Beyond entertainment, the team sees Genie 3 as a game-changer for training AI. Instead of training a self-driving car or a robot in the real world (which is slow and dangerous), you can create infinite simulations. You can even prompt rare events to happen, like a deer running across the road, to teach an AI how to handle unexpected situations safely.
The Future of Entertainment: this could lead to a YouTube version 2 or a new form of VR, where users can create and explore endless, interconnected worlds together, like the experience machine from philosophy.
While the technology is still a research prototype and not yet available to the public, it represents a monumental step towards creating true artificial worlds from the ground up.
Jack Parker Holder [Research Scientist at Google DeepMind in the Open-Endedness Team]
https://jparkerholder.github.io/
Shlomi Fruchter [Research Director, Google DeepMind]
https://shlomifruchter.github.io/
TOC:
[00:00:00] - Introduction: The Most Mind-Blowing Technology Ive Ever Seen
[00:02:30] - The Evolution from Genie 1 to Genie 2
[00:04:30] - Enter Genie 3: Photorealistic, Interactive Worlds from Text
[00:07:00] - Promptable World Events & Training Self-Driving Cars
[00:14:21] - Guest Introductions: Shlomi Fuchter & Jack Parker Holder
[00:15:08] - Core Concepts: What is a World Model?
[00:19:30] - The Challenge of Consistency in a Generated World
[00:21:15] - Context: The Neural Network Doom Simulation
[00:25:25] - How Do You Measure the Quality of a World Model?
[00:28:09] - The Vision: Using Genie to Train Advanced Robots
[00:32:21] - Open-Endedness: Human Skill and Prompting Creativity
[00:38:15] - The Future: Is This the Next YouTube or VR?
[00:42:18] - The Next Step: Multi-Agent Simulations
[00:52:51] - Limitations: Thinking, Computation, and the Sim-to-Real Gap
[00:58:07] - Conclusion & The Future of Game Engines
REFS:
World Models [David Ha, Jürgen Schmidhuber]
https://arxiv.org/abs/1803.10122
Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions [Rui Wang, Joel Lehman, Jeff Clune, Kenneth O. Stanley]
https://arxiv.org/abs/1901.01753
Questioning Representational Optimism in Deep Learning [Akarsh Kumar, Jeff Clune, Joel Lehman, Kenneth O. Stanley]
The Fractured Entangled Representation Hypothesis
https://arxiv.org/pdf/2505.11581
TRANSCRIPT:
https://app.rescript.info/public/share/Zk5tZXk6mb06yYOFh6nSja7Lg6_qZkgkuXQ-kl5AJqM Type a Sentence, Get a Playable 3D World in 3 Seconds - Shlomi Fuchter & Jack Parker-Holder](https://i.ytimg.com/vi/ekgvWeHidJs/mqdefault.jpg)
![SCHMIDHUBER: HOW WE WILL LIVE WITH AIs
SPONSOR MESSAGES:
***
CentML offers competitive pricing for GenAI model deployment, with flexible options to suit a wide range of models, from small to large-scale deployments.
https://centml.ai/pricing/
Juergen Schmidhuber — the man behind LSTMs, the 1991 linear transformer, the first generative adversarial networks, and artificial curiosity — sits down with Tim for a wide-ranging conversation about where AI came from, where it is going, and what it means for humanity.
This is the long-awaited second part of their interview, and Schmidhuber wastes no time. He opens with a provocation: the most influential invention of the twentieth century was not the transistor or the internet, but the Haber-Bosch process for synthesising fertiliser — the thing that made half of humanity possible. The twenty-first centurys equivalent, he argues, will be true artificial intelligence.
From there, the conversation traces the lineage of modern AI through Schmidhubers own work: the 1991 linear transformer (which scales linearly rather than quadratically), fast weight controllers, predictive coding, neural network distillation, and the GPU-powered deep learning revolution that his DanNet system helped ignite. He reflects on the hardware lottery, Nvidias rise, and why all of todays breakthroughs rest on algorithms invented in the previous millennium.
The middle section covers consciousness (modelled as a chunker-automatiser system), the path from AGI-as-tool to AGI-with-goals, the concept of Homo Ludens, and the geopolitics of the AI race between Europe, the US, and China. Schmidhuber then offers his characteristically optimistic take on existential risk: superintelligent AIs will be curious scientists fascinated by life, not terminators. He connects this to the Fermi paradox, speculating that Earth may be the first planet in our light cone to spawn an expanding AI bubble — a thought that carries, as he puts it, a lot of responsibility.
REFERENCES:
paper:
[00:05:45] The 1991 Linear Transformer (Unnormalized Fast Weight Controller)
https://people.idsia.ch/~juergen/FKI-147-91ocr.pdf
[00:11:00] Formal Theory of Creativity, Fun, and Intrinsic Motivation
https://arxiv.org/abs/0812.4360
[00:15:57] The Hardware Lottery
https://arxiv.org/abs/2009.06489
reference:
[00:25:00] NNAISENSE - AI for the Physical World
https://nnaisense.com/
book:
[00:35:00] Homo Ludens
https://en.wikipedia.org/wiki/Homo_Ludens
LINKS:
Full Transcript: https://app.rescript.info/share/ab733cfc4ca23500252b59b180c71f4a
Download PDF transcript: https://app.rescript.info/api/public/sessions/e20de2659d4ca617/pdf SCHMIDHUBER: HOW WE WILL LIVE WITH AIs](https://i.ytimg.com/vi/fZYUqICYCAk/mqdefault.jpg)
![An Astrophysicist Debunks the Singularity — Adam Becker
Astrophysicist Adam Becker, author of What Is Real?, joins Tim Scarfe to take apart the futures Silicon Valley keeps selling: the 2045 singularity, mind uploading, Mars colonies, and the AI apocalypse. His new book *More Everything Forever* argues these ideas are hugely influential, mostly evidence-free, and bankrolled by tech billionaires who need a story in which growth never ends.
Becker does the physics the boosters skip. Kurzweils law of accelerating returns rests on cherry-picked data, and every exponential ends. Grant Bezos his perpetual energy growth and humanity boils the oceans within a few centuries, then exhausts the observable universe in under 4,000 years. The stars are too far away, Mars dirt is poison, and the day the dinosaur-killing asteroid hit Earth was still nicer than any day on Mars. On AI, Becker calls LLMs pocket calculators for language: hallucination is the model doing exactly what it always does, and the intelligence explosion assumes intelligence is a single number you can buy with compute.
The sting is that Becker thinks the doomers are sincere. Yudkowsky, Bostrom and the effective altruists are not grifters, he says, just wrong, and their warnings that AI could end the world feed the same growth story the money depends on. He closes with his own prescription: take social problems seriously, regulate the whole tech industry, and tax billionaires out of existence.
TIMESTAMPS:
00:00:00 Cold open and the thesis of More Everything Forever
00:04:24 Kurzweils singularity and the physical limits of exponential growth
00:14:02 High agency and the fantasy of imprinting humanity on the cosmos
00:16:55 Mind uploading, functionalism, and embodied cognition
00:24:24 AI psychosis and anthropomorphizing LLMs
00:26:24 Calculators, hallucination, and the limits of scale
00:32:20 Yudkowsky and the intelligence-explosion argument
00:40:37 True believers, venture capital, and the sci-fi growth narrative
00:47:21 From Extropians to EA: utilitarianism and longtermism
00:53:50 Brain worms and Beckers prescription: take social science seriously
00:56:49 Why the AI-ethics discourse is broken
01:01:42 The eugenics and IQ argument against intelligence
01:06:07 Why space settlement fails: Mars, the moon, and orbital data centers
01:10:42 Billionaire myths and the search for purpose
01:13:38 Tax billionaires, regulate tech: closing prescriptions
REFERENCES:
book:
[00:00:07] More Everything Forever (Adam Becker, 2025)
https://www.hachettebookgroup.com/titles/adam-becker/more-everything-forever/9781541619593/
[00:00:15] What Is Real? (Adam Becker, 2018)
https://en.wikipedia.org/wiki/What_Is_Real%3F
[00:15:46] What We Owe the Future (Will MacAskill, 2022)
https://www.hachettebookgroup.com/titles/william-macaskill/what-we-owe-the-future/9781541618626/
other:
[00:00:27] Dreaming Against the Machine (podcast)
https://www.dreamingagainstthemachine.com
[00:01:04] The Useful Idiots of AI Doomsaying (Adam Becker, The Atlantic, 2025)
https://www.theatlantic.com/books/archive/2025/09/what-ais-doomers-and-utopians-have-in-common/684270/
concept:
[00:14:03] Agency (philosophy)
https://en.wikipedia.org/wiki/Agency_(philosophy)
[00:18:08] Embodied cognition
https://plato.stanford.edu/entries/embodied-cognition/
[00:19:06] Functionalism
https://plato.stanford.edu/entries/functionalism/
[00:20:46] Good regulator theorem
https://en.wikipedia.org/wiki/Good_regulator
[00:33:35] Intelligence explosion
https://en.wikipedia.org/wiki/Intelligence_explosion
[00:34:14] Instrumental convergence
https://en.wikipedia.org/wiki/Instrumental_convergence
[00:36:26] Orthogonality thesis
https://en.wikipedia.org/wiki/Orthogonality_thesis
[00:38:06] Intentional stance
https://en.wikipedia.org/wiki/Intentional_stance
[00:44:23] Effective altruism
https://en.wikipedia.org/wiki/Effective_altruism
[00:47:32] Extropianism
https://en.wikipedia.org/wiki/Extropianism
[00:49:50] Utilitarianism
https://en.wikipedia.org/wiki/Utilitarianism
[00:52:15] Longtermism
https://en.wikipedia.org/wiki/Longtermism
[00:56:02] Human biodiversity (HBD)
https://en.wikipedia.org/wiki/Human_biodiversity
[01:05:00] Recursive self-improvement
https://en.wikipedia.org/wiki/Recursive_self-improvement
[01:09:25] Speed of light
https://en.wikipedia.org/wiki/Speed_of_light
person:
[00:04:49] Ray Kurzweil
https://en.wikipedia.org/wiki/Ray_Kurzweil
[00:15:42] Will MacAskill
https://en.wikipedia.org/wiki/William_MacAskill
[00:17:37] Adrian Daub
https://en.wikipedia.org/wiki/Adrian_Daub
[00:26:05] Shannon Vallor — The AI Mirror
https://en.wikipedia.org/wiki/Shannon_Vallor
[00:33:05] Eliezer Yudkowsky
https://en.wikipedia.org/wiki/Eliezer_Yudkowsky
[00:35:21] Nick Bostrom
https://en.wikipedia.org/wiki/Nick_Bostrom
[00:50:22] Peter Singer
https://en.wikipedia.org/wiki/Peter_Singer
[00:57:45] Timnit Gebru
https://en.wikipedia.org/wiki/Timnit_Gebru
[01:04:53] I. J. Good
https://en.wikipedia.org/wiki/I._J._Good An Astrophysicist Debunks the Singularity — Adam Becker](https://i.ytimg.com/vi/fgsmq8f3sWQ/mqdefault.jpg)

![Why Frontier AI Labs Fight to Hide Chain of Thought — Ilia Shumailov & Alexander Panfilov
Tim Scarfe speaks with Ilia Shumailov and Alexander Panfilov about their paper, Stealing Reasoning Traces from Proprietary LLM APIs.
The core bug sounds deceptively simple: providers return encrypted reasoning state so conversations can be resumed or forked. But those blobs can be replayed across users and sibling models. A smaller model can ask the provider to decrypt the trace, then repeat the hidden reasoning in plain text. The discussion covers leaked private data, a broadly reusable jailbreak, poisoned agent traces, chain-of-thought monitoring, responsible disclosure, and possible defenses.
Ilia Shumailov is an AI and security researcher, formerly at Google DeepMind, who completed his Cambridge PhD under Ross Anderson. Alexander Panfilov is a PhD researcher at the ELLIS Institute Tübingen and the Max Planck Institute for Intelligent Systems, working on AI safety, adversarial machine learning, and LLM red-teaming. They close by separating the demonstrated jailbreaking threat from ordinary benign distillation, and by arguing for controlled experiments over sweeping claims.
TIMESTAMPS:
00:00:00 Intro montage
00:01:33 Portable encrypted thought and decoded reasoning
00:24:55 How the attack works and what it means
00:39:04 Doom, defense, and scientific restraint
REFERENCES:
paper:
[00:00:00] Stealing Reasoning Traces from Proprietary LLM APIs
https://arxiv.org/abs/2608.09867
[00:09:22] Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
https://arxiv.org/abs/2507.11473
[00:11:30] Reasoning Models Don’t Always Say What They Think
https://www.anthropic.com/research/reasoning-models-dont-say-think
[00:37:22] PostTrainBench: Can LLM Agents Automate LLM Post-Training?
https://arxiv.org/abs/2603.08640
[00:41:02] Large-scale online deanonymization with LLMs
https://arxiv.org/abs/2602.16800
other:
[00:09:28] OpenAI and Hugging Face partner to address security incident during model evaluation
https://openai.com/index/hugging-face-model-evaluation-security-incident/
[00:10:22] Claude, GPT, and Gemini All Struggle to Evade Monitors
https://metr.org/notes/2025-08-22-claude-gpt-gemini-struggle-evade-monitors/
tool:
[00:42:08] Isabelle proof assistant
https://isabelle.in.tum.de/
RESCRIPT:
https://app.rescript.info/share/07fc38276e0823dc9b8986c32e202c7f Why Frontier AI Labs Fight to Hide Chain of Thought — Ilia Shumailov & Alexander Panfilov](https://i.ytimg.com/vi/gasgivVCl2U/mqdefault.jpg)


![The Hidden Math Behind All Living Systems
Dr. Sanjeev Namjoshi, a machine learning engineer who recently submitted a book on Active Inference to MIT Press, discusses the theoretical foundations and practical applications of Active Inference, the Free Energy Principle (FEP), and Bayesian mechanics. He explains how these frameworks describe how biological and artificial systems maintain stability by minimizing uncertainty about their environment.
Namjoshi traces the evolution of these fields from early 2000s neuroscience research to current developments, highlighting how Active Inference provides a unified framework for perception and action through variational free energy minimization. He contrasts this with traditional machine learning approaches, emphasizing Active Inferences natural capacity for exploration and curiosity through epistemic value.
The discussion covers key technical concepts like Markov blankets,
generative models, and the distinction between continuous and discrete implementations. Namjoshi explains how Active Inference moved from continuous state-space models (2003-2013) to discrete formulations (2015-present) to better handle planning problems.
He sees Active Inference as being at a similar stage to deep learning in the early 2000s - poised for significant breakthroughs but requiring better tools and wider adoption. While acknowledging current computational challenges, he emphasizes Active Inferences potential advantages over reinforcement learning, particularly its principled approach to exploration and planning.
Namjoshi advocates for balanced oversight that enables innovation while maintaining appropriate safeguards. He expresses particular concern about the rapid pace of AI development potentially outpacing our understanding of risks and regulatory frameworks.
Dr. Sanjeev Namjoshi
https://snamjoshi.github.io/
TOC:
1. Theoretical Foundations: AI Agency and Sentience
[00:00:00] 1.1 Intro
[00:04:30] 1.2 Free Energy Principle and Active Inference Theory
[00:11:16] 1.3 Emergence and Self-Organization in Complex Systems
[00:19:11] 1.4 Agency and Representation in AI Systems
[00:29:59] 1.5 Bayesian Mechanics and Systems Modeling
2. Technical Framework: Active Inference and Free Energy
[00:38:37] 2.1 Generative Processes and Agent-Environment Modeling
[00:42:27] 2.2 Markov Blankets and System Boundaries
[00:44:30] 2.3 Bayesian Inference and Prior Distributions
[00:52:41] 2.4 Variational Free Energy Minimization Framework
[00:55:07] 2.5 VFE Optimization Techniques: Generalized Filtering vs DEM
3. Implementation and Optimization Methods
[00:58:25] 3.1 Information Theory and Free Energy Concepts
[01:05:25] 3.2 Surprise Minimization and Action in Active Inference
[01:15:58] 3.3 Evolution of Active Inference Models: Continuous to Discrete Approaches
[01:26:00] 3.4 Uncertainty Reduction and Control Systems in Active Inference
4. Safety and Regulatory Frameworks
[01:32:40] 4.1 Historical Evolution of Risk Management and Predictive Systems
[01:36:12] 4.2 Agency and Reality: Philosophical Perspectives on Models
[01:39:20] 4.3 Limitations of Symbolic AI and Current System Design
[01:46:40] 4.4 AI Safety Regulation and Corporate Governance
5. Socioeconomic Integration and Modeling
[01:52:55] 5.1 Economic Policy and Public Sentiment Modeling
[01:55:21] 5.2 Free Energy Principle: Libertarian vs Collectivist Perspectives
[01:58:53] 5.3 Regulation of Complex Socio-Technical Systems
[02:03:04] 5.4 Evolution and Current State of Active Inference Research
6. Future Directions and Applications
[02:14:26] 6.1 Active Inference Applications and Future Development
[02:22:58] 6.2 Cultural Learning and Active Inference
[02:29:19] 6.3 Hierarchical Relationship Between FEP, Active Inference, and Bayesian Mechanics
[02:33:22] 6.4 Historical Evolution of Free Energy Principle
[02:38:52] 6.5 Active Inference vs Traditional Machine Learning Approaches
Transcript and shownotes with refs and URLs:
https://www.dropbox.com/scl/fi/qj22a660cob1795ej0gbw/SanjeevShow.pdf?rlkey=w323r3e8zfsnve22caayzb17k&st=el1fdgfr&dl=0
SELECTED REFS:
[0:02:45] Fristons original Free Energy Principle paper (Nature Reviews Neuroscience, 2010) - foundational text establishing FEP
[0:35:55] Schrödingers What is Life? (1944) - pioneering work connecting physics and biology
[0:44:30] Bayes Theorem - fundamental mathematical framework underlying probabilistic inference
[0:58:25] Shannons Mathematical Theory of Communication (1948) - established information theory
[1:18:20] Parr, Pezzulo & Fristons Active Inference (MIT Press) - comprehensive synthesis of the field
[1:24:05] Kahnemans Thinking, Fast and Slow - seminal work on dual-process theory of cognition
[1:25:55] Simons concept of satisficing - fundamental contribution to bounded rationality theory
[2:23:05] Dawkins The Selfish Gene (1976) - influential evolutionary theory perspective
[2:34:15] MacKays work on information theory and machine learning - bridged information theory and modern ML The Hidden Math Behind All Living Systems](https://i.ytimg.com/vi/hf18w6CuY8o/mqdefault.jpg)