My latest "Expert Voices" column for Science Magazine:
"AI's Challenge of Understanding the World".
Melanie Mitchell
Professor at the Santa Fe Institute. Interested in the nature of intelligence.
My latest Substack post:
Can Large Language Models Reason?
https://aiguide.substack.com/p/can-large-language-models-reason
The Santa Fe Institute seeks applications for postdoctoral fellows. This is the best postdoc position in the world for interdisciplinary scientists!
More information and application link can be found at https://apply-sfi.smapply.org/prog/complexity_postdoctoral_fellowship_/
Application deadline: October 13, 2023.
Excellent piece in Tech Review on on evaluation issues for LLMs.
Article in Nature today on the question of how to assess AI systems, including some recent work by myself and my collaborators!
I'm honored & excited to give the annual Margaret Boden Lecture, sponsored by the Leverhulme Centre for the Future of Intelligence.
In person in Cambridge, UK, on August 18.
I will talk about still-remaining, all-important challenges for abstraction and analogy in AI systems.
It was an honor for me to participate in this Munk Debate on AI. If you missed it you can watch it here:
Thanks for inviting me!
Did GPT-4 hire and then lie to a Task Rabbit worker to solve a CAPTCHA?
https://aiguide.substack.com/p/did-gpt-4-hire-and-then-lie-to-a
Melanie Mitchell boosted:
New postdoc job in AI / CogSci available at the Santa Fe Institute.
Are you a grad student or postdoc interested in working with me on AI systems for abstraction and analogy?
See https://santafe.edu/about/jobs/postdoc-ai for more info. Apply by June 9.
Please share!
(View from our campus)
New postdoc job in AI / CogSci available at the Santa Fe Institute.
Are you a grad student or postdoc interested in working with me on AI systems for abstraction and analogy?
See https://santafe.edu/about/jobs/postdoc-ai for more info. Apply by June 9.
Please share!
(View from our campus)
New paper from my research group:
The ConceptARC Benchmark: Evaluating Understanding and Generalization in the ARC Domain
https://arxiv.org/abs/2305.07141
I summarize the paper in my latest Substack post:
https://aiguide.substack.com/p/on-evaluating-understanding-and-generalization
I finally got around to listening to this interview with Murray Shanahan. As expected, it's really good. I agree with almost everything Murray said.
(And BTW, Many Minds is an excellent podcast.)
And just added to the playlist:
Yann LeCun: Towards Machines That Can Understand, Reason, & Plan
https://www.youtube.com/watch?v=_JfEScYyVCE&list=PLZlVBTf7N6GpOCwMH-TIhwrUwV8YJMANM&index=9
Reposting from Twitter:
Recently I hosted a fantastic workshop at the Santa Fe Institute, entitled "AI and the Barrier of Meaning 2". Description below.
We had eminent speakers from many fields to discuss these issues, including AI, cognitive psychology, linguistics, anthropology, sociology, philosophy, complex systems, and law.
Most of the talks are now on SFI's YouTube channel. Here's a link to the entire playlist:
https://www.youtube.com/playlist?list=PLZlVBTf7N6GpOCwMH-TIhwrUwV8YJMANM
Do half of AI researchers believe that there's a 10% chance AI will kill us all?
I fact-check a widespread claim.
https://aiguide.substack.com/p/do-half-of-ai-researchers-believe
Yes, here is a preprint:
https://melaniemitchell.me/PapersContent/BurnellEtAlScience2023.pdf
It's time to rethink how we evaluate AI systems.
I'm proud to be a co-author of this new paper (led by @ryanburnell ), which has some recommendations:
My "thoughts on a crazy week in AI news".
https://aiguide.substack.com/p/thoughts-on-a-crazy-week-in-ai-news
Perspective paper by myself and David Krakauer is now published in PNAS:
"The Debate Over Understanding in AI's Large Language Models".
Client Info
Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst