cd /news/artificial-intelligence/advancing-science-together-highlight… · home › topics › artificial-intelligence › article
[ARTICLE · art-173404] src=about.fb.com ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Advancing Science Together: Highlights From the Open Source AI For Science Summit

Meta's Fundamental AI Research (FAIR) team expanded its Digital Brain Project, raising the open brain-data target from 15,000 to 25,000 hours, at the Open Source AI for Science Summit held at its Menlo Park, California headquarters. FAIR also released UMA Nitro, a faster version of its Universal Model for Atoms, and announced that Discovery Education is making the UMA Playground chemistry tool available to high school chemistry teachers, reaching more than 4 million educators and 45 million students. The University of Pittsburgh presented an assistive-robotics system built on DINOv3 and SAM 3 that it plans to demonstrate on a robot helping a person eat.

key takeaway

UMA Nitro is a faster version of Meta's Universal Model for Atoms, and UMA models identified many crystal forms of drug molecules without additional training data.

by read3 min views6 publishedNov 20, 2026
Advancing Science Together: Highlights From the Open Source AI For Science Summit
Image: About (auto-discovered)

From studying the brain to discovering new materials, open source AI is giving scientists new ways to tackle difficult problems. Last month, Meta’s Fundamental AI Research (FAIR) team brought researchers from universities, hospitals, and industry together at our Menlo Park, California, headquarters for the Open Source AI for Science Summit. Over two days, attendees explored how openly shared AI models, tools, and data can advance healthcare, neuroscience, chemistry, materials science, and mathematics. Hands-on workshops also gave attendees a chance to collaborate directly with the scientists behind Meta’s models.

Here are a few highlights.

Advancing Medical Research and Real-World Applications #

We announced an expansion of the Digital Brain Project, launched with the Rothschild Hospital Foundation and the University of Montreal. The project funds researchers to collect de-identified brain data and share it openly. We’re increasing the project’s data target from 15,000 hours of brain data to 25,000, to help researchers understand human intelligence and explore ways to improve neurological health.

Researchers are also exploring how AI can support greater independence for people with disabilities.The University of Pittsburgh presented an assistive-robotics system built on DINOv3 and SAM 3. The team plans to demonstrate a robot that can help a person eat and is experimenting with letting users give instructions in their own words.

Accelerating Chemistry and Bringing It Into Classrooms #

Scientists use computer simulations to identify promising materials and medicines before testing them in a lab. At the summit, we released UMA Nitro, a faster version of Meta’s Universal Model for Atoms.

In a study with Carnegie Mellon and UC Riverside, UMA models identified many crystal forms of drug molecules without additional training data. Knowing a drug’s crystal form helps researchers understand how a pill dissolves. With the University of Chicago, we also showed how UMA can simulate entire enzymes, atom by atom, potentially helping scientists understand how they work and engineer better ones for medicine, sustainability and energy.

The same technology can make chemistry more accessible to students. UMA Playground transforms how we visualize chemistry, letting users build molecules, watch bonds break and form, and explore chemical reactions. Discovery Education announced that the tool is now available to high school chemistry teachers, with classroom guides. RealChem, a college chemistry course from Arizona State and Carnegie Mellon, plans to integrate it into student activities. Discovery Education reaches more than 4 million educators and 45 million students, while RealChem is used by nearly 100 institutions.

Making Math a Community-Driven Science #

Checking a new mathematical proof can take experts months or even years. Lean, an open source language for writing mathematics, lets computers verify each logical step. That makes it easier for mathematicians to build on one another’s work, whether a result comes from a colleague, a student, or an AI assistant, without checking the same proof again.

This approach could enable larger, more open collaborations, like those that drive discoveries in physics and genomics. Through it all, we work hand-in-hand with the mathematics community to keep their research questions and priorities at the center. Mathematicians guide the work — they choose the questions and make sure the statements written in Lean reflect what they mean.

At the summit, FAIR researchers shared open source projects designed to support this work, including ATLAS, an open library that formalizes 26 open-access textbooks and more than 45,000 computer-checked definitions and results, giving researchers a shared foundation to build on.

By sharing models, code and research openly, we’re helping scientists and educators build on one another’s work. The summit showed how partnerships like these can extend AI’s impact beyond Meta — and create more opportunities for discovery.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @meta 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/advancing-science-to…] indexed:0 read:3min 2026-11-20 · —