Amin Karbasi

Neurips is running the largest LLM scientific evaluation in history. We just found a human in the loop.

I’m gonna be recruiting students thru both @LTIatCMU (NLP) and @CMU_EPP (Engineering and Public Policy) for fall 2026! If you are interested in reasoning, memorization, AI for science & discovery and of course privacy, u can catch me at ACL! Prospective students fill this form:

AI+Science book is now up, After years of work, our book on Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems is available at Foundations and Trends® in Machine Learning. We cover the history of works in the intersection of AI and Science, to…

If you are interested in safety/security jailbreaking of LLMs, defenses against them, and how the safety issues become more complicated when we design agentic workflows, this tutorial by @HamedSHassani, @aminkarbasi, @AlexRobey23 is highly recommended

There is also one important issue. Did IMO verify the solutions? Do they agree with the claims?

Wildly impressive. Humanity’s greatest mental exercise, mathematics, is being tackled by what is essentially glorified next-token prediction.

Ran into my younger self at @icmlconf. He looks way more rested.

On Monday, I'll be presenting a tutorial on jailbreaking LLMs + the security of AI agents with @HamedSHassani and @aminkarbasi at ICML. I'll be in Vancouver all week -- send me a DM if you'd like to chat about jailbreaking, AI agents, robots, distillation, or anything else!

My brilliant student @AnayMehrotra and amazing post-doc Alkis Kalavasis just won the Best Paper Award at COLT 2025! Clearly, they’ve cracked the code: do great math, write a killer paper, take home the trophy. Huge congratulations to both of them. Now please don’t forget your…

I think he means mathematics :)

Unfortunately, the adding information part will lead to so much hallucination that the model may not be able to recover from. @elonmusk, correctly, always talks about following the first principles, asking whether there is a fundamental reason we may be wrong. I think in this…

“High Dimensional Probability” is one of favorite books of all time. I even taught it once. Now, the second edition is out. math.uci.edu/~rvershyn/pape…

Let machines become you!

Register now (first-come first-served) for the "Math of Trustworthy ML workshop" at #LagoMaggiore, Switzerland, Oct 12-16 this year, with a great speaker lineup and the opportunity to present your work as a poster session or contributed talk. Details @ mmlworkshop.ethz.ch

Mark Twain said "history rhymes". It is not that it repeats itself, it rhymes. I would say the same about AI benchmark crisis.

Jesus was here

An amazing list of speakers.

I also have AI winter by the way

AI summer

Together with @HamedSHassani , @AlexRobey23 , and @yaronsinger , we will be presenting an @icmlconf 2025 tutorial: “Jailbreaking LLMs and Agentic Systems: Attacks, Defenses, and Evaluations.” Key takeaways: --Latest jailbreak techniques—from prompt injections to agent exploits…