There are monsters in your LLM. (Murray Shanahan) @MachineLearningStreetTalk

Uploaded July 2024 | Updated September 2026, 2 weeks ago

Murray Shanahan — professor of Cognitive Robotics at Imperial College London and senior research scientist at DeepMind — challenges how we think and talk about machine intelligence. He argues that the biggest danger of anthropomorphism is not emotional attachment but systematic misattribution of capabilities, in both directions.

The conversation spans simulators and simulacra (drawing on Janus from LessWrong), the shoggoth theory of what lies behind the RLHF mask, Wittgenstein's private language argument applied to AI consciousness, and whether concepts like agency and embodiment are necessary for genuine understanding. Murray draws on his work as scientific advisor to Ex Machina and his papers on conscious exotica to articulate why our existing vocabulary for consciousness is inadequate for these new entities.

This is a philosophically rigorous two-hour conversation that never loses sight of the engineering reality — covering everything from the Turing test to Nagel's bat to the ARC challenge, with Murray consistently pushing back on easy answers.

---
TIMESTAMPS:
00:00:00 Intro
00:05:49 Simulators and simulacra
00:11:04 The 20 questions game and simulacra stickiness
00:18:50 Murrays experience with Claude 3
00:30:04 RLHF and alignment
00:32:41 Anthropic Golden Gate Bridge experiment
00:37:05 Agency in language models
00:41:05 Embodiment and knowledge acquisition
00:57:51 ARC challenge and abstract reasoning
01:03:31 The conscious stance
01:13:58 Space of possible minds
01:17:45 Wittgenstein private language and subjectivity
01:29:58 Conscious exotica
01:33:23 Dennett and the intentional stance
01:40:58 Anthropomorphisation risks
01:46:47 Reasoning in language models
01:53:56 The Turing test revisited
02:04:41 Nagels bat and subjective experience
02:08:08 Mark Bishop idealism and Chinese Room
02:09:32 Panpsychism and consciousness

---
REFERENCES:
book:
[00:00:00] The Technological Singularity
doc.ic.ac.uk/~mpsha
[00:41:05] Embodiment and the Inner Life
doc.ic.ac.uk/~mpsha
[01:17:45] Philosophical Investigations
en.wikipedia.org/wiki/Philosophical_Investigations
[01:17:45] The Language Game
en.wikipedia.org/wiki/The_Language_Game
[01:33:23] The intentional stance
en.wikipedia.org/wiki/Intentional_stance
[01:40:58] Metaphors We Live By
en.wikipedia.org/wiki/Metaphors_We_Live_By
website:
[00:00:00] Ex Machina
en.wikipedia.org/wiki/Ex_Machina_(film)
[00:57:51] ARC-AGI Challenge
github.com/fchollet/ARC-AGI
person:
[00:00:00] Murray Shanahan academic page
doc.ic.ac.uk/~mpsha
article:
[00:05:49] Simulators article
lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators
[01:13:58] Space of possible minds
en.wikipedia.org/wiki/Aaron_Sloman
[02:08:08] Chinese Room Argument
en.wikipedia.org/wiki/Chinese_room
paper:
[00:05:49] Role play with large language models
arxiv.org/abs/2305.16367
[00:32:41] Scaling Monosemanticity
https://transformer-circuits.pub/2024/scaling-monosemanticity/
[01:29:58] Conscious Exotica
doc.ic.ac.uk/~mpsha
[02:04:41] What Is It Like to Be a Bat
en.wikipedia.org/wiki/What_Is_It_Like_to_Be_a_Bat%3F

---
LINKS:
Full Transcript: app.rescript.info/share/a752f88b40bf0658f4e8608bd9feaa34
Download PDF transcript: app.rescript.info/api/public/sessions/7faad4e03ce9fd93/pdf

Prof Murray Shanahan:
doc.ic.ac.uk/~mpsha (look at his selected publications)
scholar.google.co.uk/citations?user=00bnGpAAAAAJ&hl=en
en.wikipedia.org/wiki/Murray_Shanahan
https://x.com/mpshanahan

There are monsters in your LLM. (Murray Shanahan)

$We need AIs with PHYSICAL experience (Jeff Beck) Jeff Beck spent years studying computational neuroscience with Alexandre Pouget, Peter Latham, and Wei Ji Ma before founding Noumenal Labs. In this conversation with Tim Scarfe, he argues that language models are fundamentally limited because they manipulate symbols without the physical grounding that gives those symbols meaning. Beck walks through the Bayesian brain hypothesis — the idea that our brains maintain probabilistic models of the world, built from direct sensory experience. Language, he says, is just a thin lossy summary of that richer internal model. He points to Markus Meisters work showing that human information output runs at roughly 10 bits per second, a tiny fraction of what comes in. The conversation gets interesting around the question of whether AI systems can genuinely understand anything. Becks position is clear: until a language model produces something genuinely novel — not recombined from training data — he wont call it intelligent. The second half turns to alignment — Beck argues that you cannot separate someones beliefs from their reward function just by observing their behavior, a claim with serious implications for AI safety. SPONSOR MESSAGES: *** Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. They are hiring a Chief Engineer and ML engineers. Events in Zurich. Goto https://tufalabs.ai/ *** TIMESTAMPS: 00:00:00 Bayesian brain and neural computation foundations 00:02:00 Belief-reward entanglement and AI alignment 00:02:19 Sponsor: Tufa Labs 00:02:45 Visual cortex discovery and the Hubel-Wiesel experiment 00:05:00 Theory of mind tests and ChatGPT limitations 00:08:50 The sum and product riddle: pattern recognition vs reasoning 00:09:30 Systems engineering and object decomposition 00:12:20 Vicarious experience and grounded understanding 00:15:20 Neural coding and choice of generative model 00:17:30 Line-of-sight legibility in AI reasoning 00:18:50 Language as lossy compression of cognition 00:22:50 10 bits per second: information processing bottleneck 00:25:00 Why language models cannot substitute for understanding 00:28:00 Scientific abstraction and idealization 00:29:40 Markov blankets and system partitioning 00:33:20 Scientific realism, noise, and the limits of models 00:36:00 Free energy principle as mathematical framework 00:38:40 Black-box prediction vs legible explanation 00:41:00 Reward functions and the impossibility of alignment 00:45:00 Modeling beliefs as prerequisite for value inference REFERENCES: paper: [00:00:15] Bayesian inference in neural computation (Ma, Beck, Latham, Pouget) https://www.nature.com/articles/nn1790 [00:01:50] Noumenal Labs research paper https://arxiv.org/html/2502.13161v1 [00:05:25] LLM performance on theory of mind tasks (Kosinski) https://arxiv.org/abs/2302.02083 [00:16:25] Bayesian Mechanics (Ramstead, Sakthivadivel, Heins et al.) https://arxiv.org/abs/2205.11543 [00:17:10] Building Human-like Communicative Intelligence (Dubova) https://arxiv.org/abs/2201.02734 [00:20:45] Why do we live at 10 bits/s? (Meister, Zheng) https://www.sciencedirect.com/science/article/abs/pii/S0896627324008080 [00:28:30] Markov blankets in biological systems (Friston) https://royalsocietypublishing.org/doi/10.1098/rsif.2017.0792 [00:32:15] Critique of Gabor patches in neuroscience (Tsao) https://pmc.ncbi.nlm.nih.gov/articles/PMC9564096/ [00:35:55] MaxEnt and MaxCal principles (Presse et al.) https://journals.aps.org/rmp/abstract/10.1103/RevModPhys.85.1115 [00:40:08] Reward is Enough (Silver, Singh, Precup, Sutton) https://www.sciencedirect.com/science/article/pii/S0004370221000862 [00:45:40] Modeling Human Beliefs about AI Behavior (Lang, Forre) https://www.arxiv.org/pdf/2502.21262 website: [00:01:50] Noumenal Labs (Jeff Beck) https://www.noumenal.ai/ [00:02:19] Tufa AI Labs https://tufalabs.ai/ [00:22:25] Steven Piantadosi https://colala.berkeley.edu/people/piantadosi/ [00:23:35] Mad Libs (Stern, Price) https://en.wikipedia.org/wiki/Mad_Libs [00:31:05] Mathematical Platonism (Linnebo, SEP) https://plato.stanford.edu/entries/platonism-mathematics/ book: [00:25:25] The Brain Abstracted (Chirimuuta) https://mitpress.mit.edu/9780262548045/the-brain-abstracted/ LINKS: Full Transcript: https://app.rescript.info/share/75dfad07d8cad3ed6ccd03253aa006be Download PDF transcript: https://app.rescript.info/api/public/sessions/981caa3aeba083df/pdf Extended version on patreon: https://www.patreon.com/posts/jeff-beck-125455115$