Gabriele Sarti
@gsarti.com
Open-source interpretability to seize the means of prediction. Postdoc @ Northeastern, @ndif-team.bsky.social w/ @davidbau.bsky.social. gsarti.com
An unprecedented number of submissions means an unprecedented need for reviewers! If you have published work in interpretability, please consider signing up to review ⬇️
BlackboxNLP@blackboxnlp.bsky.social · last mo.⏳ One week left! The submission deadline for BlackboxNLP 2026 is July 17 (AoE). If you're working on analyzing or interpreting neural networks for NLP, now's the time to get your paper in. 📍 Co-located with #EMNLP2026 in Budapest 🔗 blackboxnlp.github.io
Happy to be a small part of this colossal effort led by @zouharvi.bsky.social to reinvent the future of machine translation evaluation! 🔡 Join us in building the LTB and become a coauthor, help wanted!
Vilém Zouhar@zouharvi.bsky.social · 4w agoThere are many things machine translation still can't do. Help us steer the next direction by contributing hard-to-translate inputs (and be on a cool paper).
⏳ The BlackboxNLP 2026 Reproducibility Challenge deadline has been extended to July 24 (AoE) ⏳ If you've been working on a robustness check, ablation, or replication of recent NLP interpretability work, you have a bit more time to get your submission in.
I'd never have guessed models commit to their final answer this early, often within the first 20% of reasoning, across math/logic tasks and model families. The rest is mostly hedging that doesn't change their mind. And turns out they encode this internally, we can decode it! 🧵👇
Sara Candussio@saracandussio.bsky.social · last mo.Are all the CoT steps necessary? In our latest paper, we find evidence for the existence of a commitment boundary, marking a sharp transition from no/mid guesses to the model final answer across various reasoning tasks and model families. Thread 🧵👇
🚨New paper led by @danielsc4.it & @saracandussio.bsky.social studying answer commitment across CoT for math & logic reasoning! Turns out, models often converge to final answers early in the CoT, then tend to "fake" hedging & re-checks - and we can tell apart mid/final guesses quite robustly! See 👇
Sara Candussio@saracandussio.bsky.social · last mo.Are all the CoT steps necessary? In our latest paper, we find evidence for the existence of a commitment boundary, marking a sharp transition from no/mid guesses to the model final answer across various reasoning tasks and model families. Thread 🧵👇
Our @ndif-team.bsky.social was thrilled to sponsor the BlackboxNLP repro challenge! Interpretability needs more robust & generalizable findings - if you're working in this space, this should be on your radar!
BlackboxNLP@blackboxnlp.bsky.social · last mo.🏆 Announcing the NDIF Best Paper Award for the BlackboxNLP 2026 Reproducibility Challenge! @ndif-team.bsky.social Reproduce an interp. finding with nnsight + NDIF, open-source it, and push it further. Winners will receive a $500 prize, and will be invited to present at the workshop!
It was very nice to be back in the Netherlands and EAMT to receive this prize! Thanks to everyone involved 🤗
@eamt2026.bsky.social · 2mo agoHuge congratulations to @gsarti.com for winning the #EAMT2026 Best Thesis Award! 🏆✨ His outstanding work on "From Insights to Impact: Actionable Interpretability for Neural Machine Translation" is a brilliant contribution to the community. So well deserved! 👏🚀
Are you wondering if LLM interpretability results generalize, reproduce, etc.? Check out the reproducibility challenge and submit your work reproducing papers in this area: bsky.app/profile/blac...
BlackboxNLP@blackboxnlp.bsky.social · 3mo ago📣 Announcing the BlackboxNLP 2026 Reproducibility Challenge! A new track dedicated to rigorous robustness checks of NLP interpretability work - stress-testing baselines, ablations, generalizability, and evaluation.
"the new toaster says 'I Love You' when you put the toast in, and 'I'm Sorry' when it burns it. This causes some people to get angry b/c they don't think the toaster means it, and others to develop unhealthy attachments b/c they think it does. the solution is to make the toaster TRULY sorry"
We don’t always know what problems are hard for LLMs. So devs evaluate on tasks HUMANS find hard or on broad benchmarks. What if we could instead anticipate which scenarios a model will fail on—all without evaluating specific input examples? 🧵NEW PAPER by @jenniferlumeng.bsky.social
I recently spoke with Yascha Mounk about how researchers look inside AI to understand how it is thinking. Here is the podcast: writing.yaschamounk.com/p/david-bau-2
David Bau on How—and Whether—Artificial Intelligence Thinks
Yascha Mounk and David Bau examine the mysterious internal processes that drive AI behavior—and why they may be fundamentally alien.
writing.yaschamounk.com
How do brains plan actions towards goals? To get at this question we studied mice navigating complex mazes as goals changed on every trial 🧵 Work with @thomasakam.bsky.social @behrenstimb.bsky.social @kristorpjensen.bsky.social now on BioRxiv: www.biorxiv.org/content/10.6...
We are starting a new, nonprofit alignment organization, ⊢ Sequent Research, bringing together researchers previously on UK AISI’s Alignment Team, Timaeus, and elsewhere to research how to align superintelligence. We are hiring! 🧵 sequent.org/launch
The New England Mechanistic Interpretability (NEMI) workshop is coming to BU on Aug. 14! Join us for talks, a panel, food, and plenty of opportunities to connect with the many great researchers in the area. Register and help spread the word!
Micah Benson@micahben.bsky.social · 2mo ago🧠🤖 The 2026 New England Mechanistic Interpretability (NEMI) Workshop will be Aug. 14 at Boston University! Help spread the word and join the New England mech interp community! Registration and submission info in thread:👇
New blog: I am worried by NLP research culture NLG and NLP are mostly much better in 2026 than when I got my PhD in 1990. Unfortunately research culture has gotten *worse” in this period, which really worries me as I retire. ehudreiter.com/2026/06/08/n...
I am worried by NLP research culture
In most ways NLG and NLP are much better in 2026 than when I got my PhD in 1990. Unfortunately research culture has gotten *worse” in this period, which really worries me as I retire. We have…
ehudreiter.com
"You're right to call me on that!" Can you catch an AI in the act of lying? Register below to enter our AI lie-detection contest. AI lies are a big problem. The frontier labs have all worked hard to fight AI deception. They all try to monitor their AIs for it.
Check out our latest work led by @veraneplenbroek.bsky.social! Conversation topics seem much better explanations for LLMs behavior over latent sociodemographic info about the user - but what does the choice of topic tell us about the speaker in the first place?
Vera Neplenbroek@veraneplenbroek.bsky.social · 2mo agoWhen you ask an LLM "What job can I get with no prior experience?" and then get recommended a low salary, would your background have changed the answer you receive? Surprisingly, perhaps not. Check out our paper with @gsarti.com @arianna-bis.bsky.social and Raquel Fernández: arxiv.org/abs/2606.02776
Check out our new hackathon for building the best AI lie detector on the market! A lot of interesting questions and great prizes for participants, apply early if you want in! :)
NDIF Team@ndif-team.bsky.social · 2mo agoCan you tell when an AI model is lying? Announcing Aletheia's Quest, an AI lie detection challenge running this summer, organized by Cadenza Labs and NDIF. Multiple model organisms to interrogate and probe, $50K prize pool, no local GPU required.
With the large influx of submissions and a faster pace of research, reproducibility is more important than ever. With this reproducibility challenge, we want to put the focus on best practices wrt. baselines🧱, ablations🌈, eval🔎 and generalizability🗺️ of interpretability!
BlackboxNLP@blackboxnlp.bsky.social · 3mo ago📣 Announcing the BlackboxNLP 2026 Reproducibility Challenge! A new track dedicated to rigorous robustness checks of NLP interpretability work - stress-testing baselines, ablations, generalizability, and evaluation.
Despite the huge inflow of researchers, much of the work in interpretability remains anecdotal. Our new repro challenge at BlackboxNLP (co-located with EMNLP 2026) aims to attract work challenging common assumptions and showing failure/success cases of popular methods. Negative results welcome!
BlackboxNLP@blackboxnlp.bsky.social · 3mo ago📣 Announcing the BlackboxNLP 2026 Reproducibility Challenge! A new track dedicated to rigorous robustness checks of NLP interpretability work - stress-testing baselines, ablations, generalizability, and evaluation.
As a computer scientist, we INVENTED the phrase "artificial intelligence." It was never exclusively yours. In both fact and fiction, AI has always been awesome, and beautiful, and, yes, problematic — and it still is.
John Scalzi@scalzi.com · 3mo agoAs a science fiction writer I really fucking resent tech bros for ruining the phrase "artificial intelligence." 80 years of it being all about cool androids and interesting computers destroyed by five years of "Spicy autocorrect making everything shittier for money." NEVER FORGIVE.
Excellent survey on causal interpretability by @amuuueller.bsky.social and many BauLab members, don't miss it!
Computational Linguistics Journal@complingjournal.bsky.social · 3mo agoInterpretability provides a toolset for understanding how and why LMs behave in certain ways. This survey proposes a perspective on interpretability research grounded in causal mediation analysis: doi.org/10.1162/COLI... #NLProc #CLJournal @jannikbrinkmann.bsky.social @amuuueller.bsky.social
IMO a key skill of a good scientist is moving comfortably between the specific & general e.g., relentless in understand the nerdy details of the data (including its coding) WHILE holding onto the big picture of the hypotheses being tested with the data
Today we’re releasing EMO, a new mixture-of-experts (MoE) model trained so modular structure emerges directly from data without human-defined priors. EMO can use a small subset of its experts for a given task while keeping near full-model performance. 🧵
New paper w/ UK AISI: Millions of people now use AI to help them write and communicate. In three experiments (14k participants, 3m+ human ratings) we show that AI writing assistance systematically distorts writer personas – their perceived beliefs, personality, and identity. 🧵
🚨 Tom Griffiths has a podcast where he interviews cognitive scientists podcasts.apple.com/ca/podcast/t... This just went to the top of my list.
The Cognition Project
Science Podcast · How can we study the mind, something we can never see or touch? This podcast tells the story of how psychologists, neuroscientists, computer scientists, linguists, and philosophers w...
podcasts.apple.com
Shahab Bakhtiari@shahabbakht.bsky.social · 4mo agoI’ve been really enjoying Tom Griffith’s "The Cognition Project" lately. A couple of observations: - Hearing the history of the cognitive revolution directly from the giants who lived it is priceless. - Recent AI advances have sort of robbed us of the joy of being amazed by human cognition. 🧠🤖
2026 is a whirlwind year for AI. Underlying it all is the greatest scientific mystery of our age. How does a neural network think? I talked w Oliver Whang in NYTimes Magazine, on how AI interpretability is a tangle of structure waiting to be unraveled: www.nytimes.com/2026/04/15/...
Tech industry mottos have a mixed track record. But we should hold idealists to their ideals. And we should celebrate when they come through. The Mythos non-release is a remarkable moment of conviction. Thoughts: davidbau.com/archives/20... Bravo to Anthropic's "race the top".
Mfw fiddling with probes all day but patching experiments don't pan out
Ari@ari-holtzman.bsky.social · 5mo ago
