Been Kim

Research Scientist at Google DeepMind, PhD from MIT. Make machines empower people.

[Been Kim @_beenkim]

In san francisco #colm feeling like the gif below in AI world. Giving a talk tmr at 9:45 am at https://t.co/KhgblpWGSW just after the famous @polynoamial's talk. If you are coming for Noam's talk and you want to stay since coffee isn't here yet, that's cool with me 😎.

[Been Kim @_beenkim]

@KindlyContrary @burny_tech I’m just a fan.

[Been Kim @_beenkim]

This piece by Scott Aaronson is like the sharpest knife that cuts through the messy, politicized discussions on AI, the risk, if we should be alarmed or not, why and what we should do.

If you don't want to do anything useful today, you've done something productive by reading this piece from top to bottom, word for word. https://t.co/f6wUPyrKG7

[Been Kim @_beenkim]

@Avani_Gupta__ 100%! I miss the days where I think about one thing for months and that alone. Sometimes coding, sometimes writing, words to words… while it is painful at times, there was some beauty in all that struggle.

[Been Kim @_beenkim]

If you are feeling sad about an AI (presumably) solving Navier–Stokes, you are not alone.

Drama aside, the fact that it only took a matter of days hits differently. Terry Tao’s words resonate deeply here:

"The mathematical community functions, in many ways, as a miniature version of humanity. It consists of individuals using a wide variety of different approaches, joined by core values. The most precious resources of our profession are students and ideas, and these we nurture with great care."

https://t.co/UmYnNK9hCn

[Been Kim @_beenkim]

@tobias_rees @rao2z What do you mean?

[Been Kim @_beenkim]

@rao2z Yeah I don’t know why this one made me particularly sad. Somehow it feels as if the beauty and the craft of human endeavor were taken away too quickly without any advances in human understanding. Even tho Humans are the ones who provided all the data that the model ate.

[Been Kim @_beenkim]

@FahadRabbani11 Yeah I don’t know exactly how the others were using ai but if somehow humans are meaningfully involved in the process I would have not been going thru the five stages of grief :)

[Thaddäus Wiedemer @thwiedemer]

When an LLM is underperforming, we tweak the prompt. Why not do the same for video models solving visual reasoning tasks?

Visual prompt engineering changes the start frame appearance but not the task, improving performance.

📄 https://t.co/kN6IUBUWRz
🌐 https://t.co/UPAzX1CAJm

[Been Kim @_beenkim]

I’m giving a talk tmr at PhilML workshop at 10:35am! Please note that they’ll be actual philosophers (other speakers!) at the workshop too, so come by for some spicy thinking questions for you to ponder as you travel back home... :) It’s a great way to end icml.

[Been Kim @_beenkim]

Flying to ICML ✈️ I’ll be spending the week either at the conference or at restaurants/street food stations eating at least 4 meals a day. Also giving a keynote at PhilML workshop on 11th:)

[Pulkit Verma @pulkit_verma]

The program for the #ICLR2026 Workshop "From Human Cognition to AI Reasoning" is now available. We have a fantastic lineup of talks.

🔗 https://t.co/InLGHBSR9I

Invited Speakers: @DocRachidAlami, @_beenkim, @ced_zhang

Co-Organizers: @julie_a_shah, @sarath_ssreedh, @si_tulli

[Been Kim @_beenkim]

Prompt engineering is still a black box. Why does changing X drastically change Y? Are there governing rules behind this evolution? Our new work proposes a simple way to uncover factors that might matter when refining prompts 👇

[Neha Kalibhat @NehaKalibhat]

Thrilled to share that our paper on "Interpreting and Controlling Model Behavior via Constitutions for Atomic Concept Edits" has been accepted at AISTATS 2026! 🚀🚀
Read more about how input mutations can be mapped to interpretable behavioral insights.
https://t.co/iRPRJoyAso
🧵

[Been Kim @_beenkim]

@francoisfleuret As my brain is taken over by “wrongfully accused” sentiment, I typed in my password to their website, brainlessly…

[Been Kim @_beenkim]

I got my account back! Thank you, first and foremost, to everyone—friends, GDM colleagues---who personally alerted me to this incident and retweeted that I'm hacked, as well as folks at X who helped me regain access. While this incident was terrible (I heard the scammers made huge money out of this), I feel incredibly lucky to have folks who cared♥️♥️♥️ (details of how this happened 👇)

[Been Kim @_beenkim]

@savvyRL haha What do you think? Do I really sound like Been Kim now? :) even if you text me, it's possible that scammers has my phone too. even if we can GVC, it might be an AI-generated content. We just have to meet in person, since robotics is not quite there yet. 😵‍💫😵‍💫😵‍💫

[Been Kim @_beenkim]

@ashVaswani Congrats!! 🎉

[Susan Zhang @suchenzang]

What a privilege it is to have time as your most valuable currency.

[Christopher Potts @ChrisGPotts]

Safety-oriented interpretability researchers should be focused on AI systems, not individual model artifacts. A snippet from the NeurIPS CogInterp workshop panel on Sunday: