@rohanpaul_ai https://x.com/rohanpaul_ai/status/1897730212164780132 #x-rohanpaul_ai
LLMs struggle with abstract concepts despite strong performance on verbal analogies.
This is because their internal representations are not truly conceptual. This paper introduces Concept Vectors (CV...
Key-value memory is an important concept in modern machine learning (e.g., transformers). Ila Fiete, Kazuki Irie, and I have written a paper showing how key-value memory provides a way of thinking about memory organization in the brain:
arxiv.org/abs/2501.02950
Excited to announce that I'll be starting a lab at the University of Montreal (psychology) and Mila (Montreal Institute of Learning Algorithms) starting summer 2025. More info to come soon, but I'll be recruiting at the Masters and PhD levels. Please share / get in touch if you're interested!
New paper from Martha Lewis and me:
"Evaluating the Robustness of Analogical Reasoning in Large Language Models"
Preprint:
arxiv.org/pdf/2411.14215
This is a much-extended follow-up on our earlier pre-print on "counterfactual tasks" in letter-string analogies.
🧵
Sixth (and last) episode of our "Nature of Intelligence" podcast is out!
This time I'm in the hot seat -- Abha interviews me about "AI's changing seasons" and lots of other AI-related stuff.
www.santafe.edu/culture/podc...
Excellent post discussing whether "AI progress is slowing down".
www.aisnakeoil.com/p/is-ai-prog...
And if you're not subscribed to @randomwalker.bsky.social and @sayash.bsky.social 's great newsletter, what are you waiting for?
Of course it's not just brute-force compute -- it's massive "heuristic search". Still defeats original purpose IMO.
Excellent post about the recent OpenAI o3 results on ARC (& other benchmarks). I don't know how @natolambert.bsky.social manages to write these so quickly! I highly recommend his newsletter.
www.interconnects.ai/p/openais-o3...
I am (more slowly) writing my own take on all this, coming soon.
Some of my thoughts on OpenAI's o3 and the ARC-AGI benchmark
aiguide.substack.com/p/did-openai...
The Society for the Neuroscience of Creativity (@tsfnc.bsky.social) has a starter pack of folks doing creativity research. We're trying to build the creativity research community here. Please follow and let us know if you'd like to be added.
Speaking of Paris... Our abstract submissions are now open! SfNC members are welcome to submit for our May 2025 conference on our new member website: www.tsfnc.org/sfnc2025-cre...
We are also accepting nominations for our 2025 awards! Find out more here: www.tsfnc.org/sfnc-awards
Interesting paper by @jaanaru.bsky.social on human vs AI creativity. Argues while both can produce similar creative products, the internal processes are very different - from neural architecture to the role of conscious experience.
How does AI impact our sense of agency in the creative process? This paper suggests a few key factors: whether AI competes or complements human abilities, how effective we perceive AI to be, and whether the stakes are high or low; AI literacy may moderate these effects.
A comprehensive review of research on creativity in AI---from artistic creation to scientific discovery---grounded in creativity theories.
Who is more creative: two humans, or a human and an AI? This study finds human teams generate better ideas than human-AI pairs, pointing to benefits of human collaboration over AI co-creativity.
New content: Ionescu, T., Goldstone, R. L., Rogobete, D., & Taranu, M. (2024). Is Cognitive Flexibility Equivalent to Shifting? Investigating Cognitive Flexibility in Multiple Domains. Journal of Cognition, 7(1): 73, pp. 1–21. DOI: doi.org/10.5334/joc.... #psychscisky
We have an opening for a PhD on creating simulation models guiding the selection of treatments for mental disorders at the University of Amsterdam with the NSMD consortium:
www.academictransfer.com/en/347555/ph...
This new format will supplement the existing 2 page submission track: Long papers are submitted earlier to accommodate proper reviews. Papers that are rejected but are within scope can transfer to the 2 page track. Of course authors can also just submit a 2 page paper directly.
After a great conference in Boston, CCN is going to take place in Amsterdam in 2025! To help the exchange of ideas between #neuroscience, cognitive science, and #AI, CCN will for the first time have full length paper submissions (alongside the established 2 pagers)! Info below👇
#NeuroAI #CompNeuro
Announcing the launch of the Open Encyclopedia of Cognitive Science: oecs.mit.edu! OECS is a freely-available, growing collection of peer-reviewed articles introducing key topics in cogsci to a broad audience of students and scholars.
Postdoc Job Alert: Help us study children's cognitive development to improve AI reasoning. vacatures.uva.nl/UvA/job/Post...
Pencil is to pencil case as sweater is to …? Since a sweater is stored in a “closet” this is what most people would choose. Young children often use association, answering “pants”, rather than reasoning with relations. LLMs seem to use association too.
Do large language models solve verbal analogies like children do? Yes, they (kind of) do! arxiv.org/abs/2310.20384
