Usman Gohar
292 posts
Usman Gohar
@UsmanGohar
Ph.D. student @IowaStateU ML Fairness, AI/Software Safety, AI Ethics
Ames, Iowa Katılım Mayıs 2010
573 Takip Edilen173 Takipçiler
A useful piece of advice I once received about writing for policymakers: youtube.com/watch?v=qDXJX1…
YouTube
English
English
Nice write up + data visualization of our recent research on AI exposure and adaptive capacity on the @washingtonpost front page today!
A really careful contextualization of findings here with other work on AI automation in the piece. h/t to @ShiraOvide and @kevinschaul
English
Have you been running evals? Consider submitting to our shared task at ACL and help unify reporting standards and transparency 🚀
EvalEval Coalition@evaluatingevals
🧪 Your LLM evaluation results could help the whole field 🚀 🧑🔬 Our ACL Shared task is out! We’re building a unified, crowdsourced database to create a common language for AI evaluation reporting. And we need your data. (1/2) evalevalai.com/events/shared-…
English
Gear up for San Diego at ACL 2026. Super excited to bring @evaluatingevals to @aclmeeting this year!! Call for papers is out 📝
EvalEval Coalition@evaluatingevals
🚨 The next edition of EvalEval Workshop is coming to @aclmeeting 2026! 🧠 Workshop on "AI Evaluation in Practice: Bridging Research, Development, and Real-World Impact" 🎇 📢 CFP is now open!!! More details ⏬ 📍 San Diego 📝 Submission deadline: Mar 12, 2026
English
Usman Gohar retweetledi
This is a great report that provides a thoughtful, detailed and very well researched description of the risks of AI. It is essential reading for anyone who wants to write or talk about AI risks.
Yoshua Bengio@Yoshua_Bengio
Today we’re releasing the International AI Safety Report 2026: the most comprehensive evidence-based assessment of AI capabilities, emerging risks, and safety measures to date. 🧵 (1/17)
English
Usman Gohar retweetledi
Writing Group
@RishiBommasani, @StephenLCasper, @TomDavidsonX, @raymondadouglas, @DavidDuvenaud, @UsmanGohar
Rose Hadshar, @ansonwhho, @tiancheng_hu, @sayashk, @Dr_Atoosa, @sj_manning, Cameron Jones, @mavroudisv , @JessicaH_Newman, Kwan Yee Ng, @RMoulange, @prpaskov, @girishsastry,
@ea_seger, @Scott_R_Singer, @charlotte_stix, @_LuciaVelasco, @nwheeler443
Advisers to the Chair
@privitera_ and @sorenmind
Senior Advisers @DAcemogluMIT @conitzer @tdietterich @FredrikHeintz @geoffreyhinton @LboroVC @susanleavy Teresa Ludermir @VidushiMarda @HelenMargetts @McDofYork @jane_munga @random_walker @AlondraNelson46, @ClaraNeppel @gramchurn Stuart Russell @MarietjeSchaake @bschoelkopf Alvaro Soto Lee Tiedrich @GaelVaroquaux Andrew Yao @yaqinzhang
English
Usman Gohar retweetledi
Today we’re releasing the International AI Safety Report 2026: the most comprehensive evidence-based assessment of AI capabilities, emerging risks, and safety measures to date. 🧵
(1/17)
English
So grateful to be part of this team. Special thanks to
@Yoshua_Bengio for chairing the initiative, the lead writers @stephenclare_,@carinaprunkl, chapter leads @ben_s_bucknall, @maksym_andr, @malcmur, the Expert Advisory Panel, @UN, @EU_Commission, @OECD & dozens of reviewers
English
The level of scientific rigor this work has undergone over the past months is exemplary, with many rounds of scrutiny from dozens of experts and stakeholders. This is reflected in the endorsements this work has received ✨
English
🚨We are finally releasing the International AI Safety Report 2026 🥳
Incredibly proud to be part of this team. I contributed as a section lead on AI-generated content and its harms. This report is the result of the cumulative efforts of dozens of experts across the world 🌎
Yoshua Bengio@Yoshua_Bengio
Today we’re releasing the International AI Safety Report 2026: the most comprehensive evidence-based assessment of AI capabilities, emerging risks, and safety measures to date. 🧵 (1/17)
English
Usman Gohar retweetledi
If you missed out on the fantastic conversations we had today, @evaluatingevals is throwing one last happy hour before we all leave San Diego 💔
Come hang out with us and discuss how evals are broken and how we as a community can do better 🤗
partiful.com/e/zn8tz8e0Qt8l…
English
We are live!! Don’t miss out on some amazing panels and talks on the world of AI evaluations
EvalEval Coalition@evaluatingevals
EvalEval is back! Our view today for the 2025 EvalEval Workshop at the beautiful @UCSD campus. We have an exciting program planned, full of wonderful discussions and people on all things evals Agenda: evalevalai.com/events/worksho… Can't be here? Join us live: meet.google.com/ozx-dsnz-gcr?h…
English
@hvngo2002 @datologyai I broke two of them (it took a beating with all the overstimulation). That said, I might have an extra one :))
English
does anyone have an extra @datologyai metal fidget spinner i lost mine 😭😔 it’s single handedly helped me deal with overstimulation at neurips
English
Operation cross the train! Didn’t need any compute or training. Can your LLM do that? Attendees after waiting for 15 mins to cross the train #NeurIPS2025
English
Absolutely loved @zeynep’s keynote. I have been thinking about the parallels to big transformations in history and she nailed it! @NeurIPSConf
English
This week, our spotlight series brings to you "The Flaw of Averages: Quantifying Uniformity of Performance on Benchmarks." The fairness community (and my research) has long discussed the problems with aggregation, but current AI Evals still remain flawed
EvalEval Coalition@evaluatingevals
✨ Weekly AI Evaluation Paper Spotlight ✨ What if the average performance scores we trust are actually hiding a benchmark’s flaws? 📰“The Flaw of Averages: Quantifying Uniformity of Performance on Benchmarks” (@aardauzunoglu, @tli104, @DanielKhashabi) introduces HARMONY. 1/n
English
You don’t want to miss this! Reach out if you have any questions :))
EvalEval Coalition@evaluatingevals
If you're at NeurIPS, you don't want to miss out!! We have an amazing program and line-up of speakers and panelists! ⏰Last week to submit your abstracts! See more: evalevalai.com/events/worksho…
English
Guess now we know why there were record submissions 😮💨
Micah Goldblum@micahgoldblum
An LLM-generated paper is in the top 17% of ICLR submissions in terms of average reviewer score, having received two 8's. The paper has tons of BS jargon and hallucinated references. Fortunately, one reviewer actually looked at the paper and gave it a zero. 1/3
English