John W. Krakauer (@blamlab) profile picture

Brain, Learning, Animation, & Movement Lab: studying motor control, motor recovery, aesthetics of action, the cognitive-motor interface, & philosophy of science

Johns Hopkins; Santa Fe Inst.
Joined February 2013
🎧 Why Can't AI Laugh at a Joke? w/ @blamlab & @adamsafron on John Krakauer's recent papers with David Krakauer and @MelMitchell1 of @sfiscience on how we might find genuine signs of emergence in LLMs, and @quiltydunn on "the deliberation taboo" in cognitive science. Full show notes & MANY more deep discussions on tech, science, and philosophy @ humansontheloop.com/p/h-41 Chapters: 00:00 Teaser & Intro 08:28 Meet John Krakauer 10:53 Against Internal Models 13:01 Replay and RL Critique 16:37 Mental (Not Neural) Representation 20:55 What Counts as Evidence 24:03 Why Humans Deliberate 26:56 #AI Mirrors the Taboo 28:42 Deliberation Taboo Framework 32:04 Chain of Thought Confabulation 34:30 Slow #Science and Creativity 38:15 #Cognition vs #Consciousness History 43:57 Pain and Conscious Advantage 46:09 Illusionism and @danieldennett 49:12 Working Memory Bottleneck Debate 51:03 @maxsbennett on Simulation Limits 53:19 Defining Real Emergence 56:38 Inside LLM Representations 57:56 NeuroAI Chick Experiments 01:04:08 Efficiency And Evolutionary Pressure 01:07:00 Laughter, Feelings, Deliberation 01:13:53 Basal, Cortical, & Conscious Intelligence 01:19:57 #WorldModels vs World Policies 01:23:27 Language’s Role & The Meaning Gap 01:27:54 Tacit Learning & One-Shot Rules 01:30:18 Admitting The Deliberation Taboo 01:31:12 Explanatory Gaps And Coarse Graining
1
3
9
902
John W. Krakauer retweeted
Most people favor one hand, and that hand tends to be the better one at writing, at throwing, at managing chopsticks. The long-standing view is that the dominant hand is “born” more capable, its skills rooted in a brain hemisphere specialized for motor control.  A new study in PNAS argues that this difference in skill is not innate at all. It is the consequence of a lifetime of practice, and it surfaces only when we pick up a tool. A reminder that even a trait as seemingly fixed as handedness can be emergent, arising from the interplay of biology, behaviour, and culture rather than from any single cause. santafe.edu/news-center/news…
2
21
4
97
7,831
John W. Krakauer retweeted
🚨New preprint! Using spinal cord stimulation, we demonstrate that non-monosynaptic corticospinal tract (CST) projections can be leveraged to enable dexterous recovery post-stroke, regardless of MEP status (a traditional measure of CST integrity). doi.org/10.64898/2026.03.24.…
1
6
16
1,612
John W. Krakauer retweeted
SFI President David Krakauer joins The Generalist Podcast to talk about intelligence, the influence of AI on human thinking, and how to preserve creativity and understanding in a rapidly changing information landscape. Watch the full interview: youtube.com/watch?v=SSBLQyi1…
This week on The Generalist Podcast, I’m interviewing David Krakauer, President of the Santa Fe Institute. Maintaining Human Intelligence in the AI Era David Krakauer is a leading complex systems researcher and the president of the @sfiscience, a unique institution dedicated to studying complex systems across disciplines. In this episode, David challenges conventional wisdom about AI, arguing that large language models pose a more immediate threat to humanity than commonly discussed existential risks—not by destroying us directly, but by eroding our cognitive capabilities through addictive, low-quality information. Listen now: • YouTube: youtube.com/watch?v=SSBLQyi1… • Spotify: open.spotify.com/episode/6Fz… • Apple: podcasts.apple.com/us/podcas… A big thank you to the incredible sponsors that make the podcast possible: ✨ Brex – The banking solution for startups: brex.com/mario ✨ Enterpret – Transform feedback chaos into actionable customer intelligence: enterpret.com/mario ✨ Persona – Trusted identity verification for any use case: withpersona.com/generalist We explore: → Why David believes LLMs aren't intelligent at all and how the AI community misunderstands emergence → The three dimensions of intelligence: inference, representation, and strategy—and which one LLMs lack → How AI acts as a "competitive" rather than "complementary" cognitive technology, atrophying our thinking abilities → What makes great minds unique, from analogical reasoning to the cultivation of unconscious creativity → How Cormac McCarthy's approach to knowledge and creativity offers lessons for the AI age → Why David believes the greatest threat from AI isn't existential risk but cognitive atrophy—"like sugar cocaine" → How to protect your mind against AI's addictive pull and maintain cognitive autonomy
2
5
1
30
10,286
John W. Krakauer retweeted
In the cerebellum, neurons are placed in competition with one another, nulling the downstream effects of each other's spikes, in a well-organized subtractive process that produces a fast changing, precise output from a collection of slow, noisy neurons.
1
24
160
8,805
John W. Krakauer retweeted
Some of my thoughts on OpenAI's o3 and the ARC-AGI benchmark: ⬇️
17
50
9
324
69,023
John W. Krakauer retweeted
Integrating brainstem and cortical functional architectures | nature.com/articles/s41593-0… led by @JustineYHansen in @NatureNeuro 400 cortical regions + 58 brainstem nuclei. Lesgo 🤘🕶️⤵️
7
110
10
296
58,859
John W. Krakauer retweeted
A recent study examining the kinematics of 3D arm movements in individuals in the sub-acute phase after #stroke found that both weakness and synergy intrusion contribute to motor deficits during the sub-acute period. journals.sagepub.com/doi/ful… #neurotwitter #neurorehabilitation
1
7
9
2,022
We just released @fchollet keynote talk from #AGI24. He challenges LLM hype, discusses ARC-AGI benchmark, and proposes merging deep learning with program synthesis for true AGI progress.
19
118
15
550
178,340
12
292
12
4,232
177,004
John W. Krakauer retweeted
Do bumblebees prefer continuous over interrupted strings to get to an artificial flower with nectar? Yes ... but does this mean that they understand the physics of a continuous string...? Find out in our latest paper in eLife: elifesciences.org/articles/9…
7
64
7
265
26,144
John W. Krakauer retweeted
The CoT hype balloon just exploded. (And kudos to @rao2z’s group for all their outstanding work on LLMs.)
6
22
9
128
30,480
Join us at the cross-sector workshop "Advances in Digital Neurorehabilitation" @unibirmingham on Sept 30th 2024! Funded @ncode_network & @LES_UniBham. With @uhammerbeck, @blamlab, Naveed Ejaz, @Mazaheri_lab, David Punt, @ChiouChloe,@Tnightingale10 et al. shorturl.at/LBh6F
1
6
3
15
2,823
John W. Krakauer retweeted
Have you heard about the million dollar ARC Prize? It's meant to get AI closer to "AGI". In my most recent blog post I wrote about the prize, how the current top programs work, and what it all might mean. Check it out (link in the next post).
11
22
4
169
15,542
The scale of the microscopic. 📽: MetaBallStudios
47
1,723
93
6,588
537,385
John W. Krakauer retweeted
I wrote about the latest attempt to sell genetic IQ prediction and why it is snake oil. A short 🧵:
28
266
45
1,184
442,883
John W. Krakauer retweeted
“I have become the person I was meant to be.” #QuadGods, an @HBO Original Documentary following the world’s first all-quadriplegic e-sports team as they navigate New York City, confront ableism, and pursue their dreams to compete as athletes, premieres July 10 on @StreamOnMax.
4
28
9
83
30,741