Log inSign up
user avatar
Been Kim
@_beenkim
Research Scientist at Google DeepMind, PhD from MIT. Make machines empower people.
people.csail.mit.edu/beenkim/
Joined August 2011
511
Following
27.3K
Followers
  • user avatar
    Been Kim
    @_beenkim
    Apr 24, 2022
    I've never imagined myself to be giving a keynote at one of the most prestigious ML conferences: ICLR. I thought those speaker seats are reserved for non-Asian, non-female, non-immigrant person who did not grow up in public housing and who has no imposter syndrome. 1/n
  • user avatar
    Been Kim
    @_beenkim
    Dec 2, 2021
    I'm hiring an intern at Google brain team 2022! Email meย if you areย 1) Graduating in 2022 (preferred) or early 2023 2) Interested in interpretability, representation learning or studying fundamental principles of how NN perceives the world. Pleas read the whole thread--> 1/n
  • user avatar
    Been Kim
    @_beenkim
    Dec 12, 2024
    Congratulations to the amazing @DGukesh โค๏ธ๐Ÿ˜๐Ÿฅฐ
    user avatar
    Gukesh D
    @DGukesh
    Dec 12, 2024
    Thanks a lot sir for all the support and encouragement๐Ÿ˜Š๐Ÿ™
    31K
  • user avatar
    Been Kim
    @_beenkim
    Nov 20, 2022
    Our paper on understanding AlphaZeroโ™Ÿ๏ธis now published at PNAS! pnas.org/doi/10.1073/pnโ€ฆ The paper "studies" AZ's internals and its behaviors in collaborations with @DeepMind and world chess champion @32gcfhkmm. What did we learn? ๐Ÿงต
    Image
    pnas.org
    Acquisition of chess knowledge in AlphaZero | PNAS
    We analyze the knowledge acquired by AlphaZero, a neural network engine that learns chess solely by playing against itself yet becomes capable of o...
  • user avatar
    Been Kim
    @_beenkim
    Oct 24, 2023
    Better way to do interpretability:โ™Ÿ๏ธInterpretability has been my passion for more than a decade. Most of time however, I was frustrated; many method don't seem to meet their promise, some even provably wrong*. I felt stuck in this impossible task.
    Image
    78K
  • user avatar
    Been Kim
    @_beenkim
    Mar 6, 2025
    ๐Ÿ”ฅ๐Ÿ”ฅOur small team in Seattle Google DeepMind is hiring! ๐Ÿ”ฅ๐Ÿ”ฅIf you are willing to move to/already in Seattle, has done significant work on human-machine communication / interpretability (from ML side) with a relevant PhD and great publication record, Join us. Apply here ๐Ÿ‘‰๐Ÿ‘‰
    57K
  • user avatar
    Been Kim
    @_beenkim
    Nov 19, 2021
    Life achievement unlocked: coauthoring paper with the founder of @DeepMind @demishassabis and world chess champion @32gcfhkmm and many other amazing folks at @DeepMind. We used high level human chess concepts to look deeper into what self-taught super-human chess player.
    user avatar
    Google DeepMind
    @GoogleDeepMind
    Nov 19, 2021
    Replying to @GoogleDeepMind
    Work by @banburismus_, @weballergy, @demishassabis, @ulrichpaquet, together with @32gcfhkmm and Andrei Kapishnikov, @adamrpearce and @_beenkim (Google Brain) 2/2
  • user avatar
    Been Kim
    @_beenkim
    Jan 11, 2024
    Many previous work of mine and others hinted โ€˜something fishyโ€™ about saliency-based methods. But we never had a rigorous proof of what we saw. This work โ€œImpossibility Theorems for Feature Attribution", now published in PNAS, to me marks a point of new beginnings.
    Image
    Image
    user avatar
    Blair Bilodeau
    @blairbilodeau
    Jan 11, 2024
    Excited to finally share that "Impossibility Theorems for Feature Attribution" is published in PNAS. TL;DR Methods like SHAP and IG can provably fail to beat random guessing. w/ @natashajaques @PangWeiKoh @_beenkim PNAS: pnas.org/doi/10.1073/pnโ€ฆ arXiv: arxiv.org/abs/2212.11870
    84K
  • user avatar
    Been Kim
    @_beenkim
    Jun 9, 2023
    Feature visualizations are widely used interpretability tools - but can we trust them? We investigate this question from an adversarial ๐Ÿฅท, empirical ๐Ÿ”ฌ and theoretical ๐Ÿ“ perspective. The result: Donโ€™t trust your eyes! (1/6) Paper: arxiv.org/abs/2306.04719 ๐Ÿงต
    Image
    88K
  • user avatar
    Been Kim
    @_beenkim
    Oct 13, 2022
    Being a Senior PC for one of the biggest conferences #ICLR2023 so far offered me 1) so many opportunities to make mistakes no matter how much time and efforts I pour in 2) to feel responsible for everything that went sub-optimally even if they were out of my control, but..๐Ÿงต
  • user avatar
    Been Kim
    @_beenkim
    May 8, 2024
    Congratulations @wellingmax and @dpkingma again for winning the first ever test of time award at ICLR! #ICLR2024 I had the honor of being the photographer:)
    Image
    26K
  • user avatar
    Been Kim
    @_beenkim
    May 6, 2023
    Now that concludes #ICLR2023 - the first major ML conference in Africa that I get to be part of as a senior PC - the job that I probably spent the *most* of my last year thinking about, losing sleep over and passed many other opportunities for. 1/3
    Image
    67K
  • user avatar
    Been Kim
    @_beenkim
    Jul 15, 2022
    Thanks everyone for waiting patiently! #iclr2023 Call for paper for ICLR 2023 is out! Full paper deadline: Sept 28 iclr.cc/Conferences/20โ€ฆ
  • user avatar
    Been Kim
    @_beenkim
    Sep 15, 2020
    Excited and honored to give keynote at #ecmlpkdd2020 tomorrow! First talk since I became a mom. So it might mostly contain my baby's pictures, and a little bit about inqterpretability.๐Ÿ˜Š