


Head of AI @AsteraInstitute Prev: AGI @DeepMind, cofounder @vicariousai (acqd by Alphabet), cofounder @Numenta. IIT-Bombay, MS&PhD Stanford. agicomics.net


Apply for an Astera Residency!
Applications are open for Astera Residencies, now through January 7, 2027! Up to $2M in funding, based out of our Emeryville hub. This round, we’re sharing a few specific project requests. Read more and apply here: astera.org/residency/ Reshare with folks who should see this!

Strongly recommend this read and sitting with the discomfort
In my ongoing quest to find the limits of Opus 5.5, I directed it to audit my old scientific field, theoretical astrophysics. We indexed about 150,000 papers, audited 50,000 for issues, and did progressively more thorough checks on thousands more. We found nearly 1000 major errors that got cited many times, most are still uncorrected. This survey enabled me to estimate the efficacy of peer review for correcting errors (not very good) and to identify dozens of open questions that anyone with spare Claude Max tokens could answer today (.md context files included). Bottom line: AI can speed up quantitative science by >1000x with no further model improvement required. But I'm seeing no evidence of this being pushed within universities, only academics complaining about labs dropping math proofs. Among my fellow reformed academic high agency founders, however, several are speed running entire disciplines. This survey has implications for how science will be run in the very near future. Hint: Academic publishing is dead. It was already a very stupid way to record and promulgate knowledge, and had been so for decades, even before all that publicly funded knowledge was locked behind paywall cartels. Now AIs can generate thousands of novel, correct papers per day, it is no longer meaningful for humans to consider a research paper a "unit of science" or to even bother much with the 99.9% of science which, up to this point, was necessary drudgery. Time to discover something like General Relativity or Evolution every week. LFG!! Enjoy: science.caseyhandmer.com/phy…

We funded 10 grad students to publish their hidden science. The work fell into three groups: results with no home in a paper, datasets too valuable to sit on a server, and tools that help other scientists work. See the results ➡ asterainstitute.substack.com…

Do these look like recursive self improvement or repetitive slop injection?

Are you an entrepreneurial scientist who has a high-risk, high-impact idea with potential for public good? Apply for the residency program at @AsteraInstitute ! Applications due January 7th. 👇
Applications are open for Astera Residencies, now through January 7, 2027! Up to $2M in funding, based out of our Emeryville hub. This round, we’re sharing a few specific project requests. Read more and apply here: astera.org/residency/ Reshare with folks who should see this!

Astera Institute Residency applications are open. Very unique opportunity to build your project in a stimulating and supportive environment! Check it out, and tell your friends about it!
Applications are open for Astera Residencies, now through January 7, 2027! Up to $2M in funding, based out of our Emeryville hub. This round, we’re sharing a few specific project requests. Read more and apply here: astera.org/residency/ Reshare with folks who should see this!

This
absolute fanfic. AI has real, immediate, tangible dangers that we need to focus on, for instance the impact on the job market and making sure random script kiddie can't use it for cyber offense. But this "oh once it's superintelligent it will take over and ensave us, maybe kill us all!" is pure sci-fi fanfic and distracts from the real (and difficult) issues, which linear probing won't solve. It's worth repeating the LLMs don't just "decide" to do something and then just "do it" in a void. They generate tokens based on our prompts, only as long as we call `model.forward()`, and on giant clusters with terabytes of weights, that's impossible to "quickly self-replicate". We also need to regularly remind that the OAI/HF incident was specifically a _guardrail-free_ cyber eval where the model is _instructed_ to hack, to find out how good it is at it. Some recent personal event made me realize there needs to be more push-back on this doomsday nonsense, and we need to re-focus the discussion on real dangers/issues. I won't argue much in the replies though, not interested in peer-reviewing Terminator fanfic.

All we now need is GOOGL 🚀
TPUs booming Gemini 4 rising Waymos buzzing Search AI-pilled Youtube GOAT-ed

on my first ever Carbon prompt it solved a research problem we’d been throwing wave after wave of Astras and Fables at without success impressed
It's been fun to pit Fable 5.2, Astra, and Hydrogen against one another on a complex scientific problem (implementing REML within our boldtailor package). I had Astra create an overview of the development process, which highlights a few things: - today's best models still pass off incorrect code as complete, even when using a test-driven development workflow - adversarial testing across models helps identify errors, but can require multiple rounds - expert judgment in the loop remains important to keep development on track github.com/poldrack/boldtail…

Excited to announce I recently joined the @AsteraInstitute to work with Dileep and his stellar NeuroAI team! We're working on the foundations of intelligence, human and otherwise. If you're at #SfN26 and interested in joining, chat with (or DM) my colleague Weinan Sun @sunw37
Dileep George @dileeplearning Feb 25 Article In search of the mystery of the cortical column and human-like general intelligence Today, Astera Institute announced that I am leading its AGI research program with an ambitious research agenda. It’s backed by $1B+ in long-term financial support for building NeuroAI that learns,


Weinan Sun @sunw37 will be at the @AsteraNeuro social at SfN to represent NeuroAI work at @AsteraInstitute. Be sure to chat with him! Astera NeuroAI will be hiring 10 postdoc researchers over the next year. Opening summer internship positions too! Keep us on your roadmap!
At #SfN26? Join @doristsao and the @AsteraNeuro team for drinks, small bites and neuroscience. 🗓 Mon, Nov 16, 8 to 10 pm 📍 Succotash Prime, 915 F St NW, a 10-minute walk from the convention center No reservation required. Bring a colleague, and please RT.

Big day here at @AsteraInstitute! We're opening applications (due Jan 7) for the next call of residents. This is across all our focus areas including life sciences, deep tech, neuroscience, AI research, and open science. See below to learn about the fully funded 12-18 month opportunity. Alongside specific navigation for projects with commercial avenues for impact, our general expectation is to work in the open. To learn about why we work this way and our open science policy see here: astera.org/open-science/ We're also looking for three new full time hires within the open science team including: An Open Science Operations Lead to work with me on all things including our internal open science efforts and external facing opportunities jobs.ashbyhq.com/astera/08fb… A Program Manager for our Federal Science Partnerships located in Washington, DC to work with our policy lead on our growing federal partnership work jobs.ashbyhq.com/astera/e1de… A Member of Technical Staff for Open Science who can serve as a data steward to help our hires and awardees generate first class data products for reusable science jobs.ashbyhq.com/astera/e26a… Apply or share - look forward to hearing from you!
Applications are open for Astera Residencies, now through January 7, 2027! Up to $2M in funding, based out of our Emeryville hub. This round, we’re sharing a few specific project requests. Read more and apply here: astera.org/residency/ Reshare with folks who should see this!

"abandon behavior cloning" is not listed because behavior cloning is used with abandon.
“DON’T WORK ON LLMs” - @ylecun at CoRL

and if you are curious about how my PhD thesis managed to corrupt him, it's here: dileeplearning.github.io/upl… :-)