FirstPrinciples

FirstPrinciples is pleased to sponsor the Machine Learning and the Physical Sciences (ML4PS) workshop at NeurIPS 2025. 🤝 For nearly a decade, ML4PS has served as a meeting ground for researchers tackling “big science” challenges. This year’s workshop and panels take on an increasingly urgent question. How do we help sustain open, curiosity-driven research in fundamental science when traditional funding models are under strain? Furthermore, what role can industry play in supporting fundamental science, and how can the insights emerging from basic research in the physical sciences and ML help catalyze wider innovations in industry and beyond? We’re proud to support a community that is pushing the boundaries of both disciplines while keeping scientific curiosity and openness at the center. Learn more: https://lnkd.in/g8--a6QS #AI #ML #AIforScience #NeurIPS

Interpreting neural networks remains challenging, largely due to their dense parametrization, global coupling of parameters, and the polysemantic behavior of neurons. Next week, Fabian Ruehle - Assistant Professor, Department of Mathematics and Physics at Northeastern University and Senior Investigator for the NSF AI Institute for Artificial Intelligence and Fundamental Interactions (IAIFI) - joins FirstPrinciples Talks to explore why Kolmogorov–Arnold Networks (KANs) behave differently. 🗓️ Register to join us! #AI #Webinar #AIforScience

Two members of the FirstPrinciples team, Alex Tessier (Director of Engineering) and Randy Davila (Member of Technical Staff, Engineering) will be at NeurIPS 2025 next week. If you’re attending and interested in discussing AI for science, collaboration opportunities, or the AI Physicist, we’d love to connect! #AIforScience #NeurIPS

Next week for FirstPrinciples Talks, we’re diving into one of the fastest-moving frontiers in scientific AI: agents that generate new research ideas. Tom Hope, research scientist at Ai2 and Head of AI/NLP research lab at The Hebrew University of Jerusalem will explore systems designed to help researchers navigate and build on the scientific literature, including: 🔹 Scimon: which explores LLMs for scientific hypothesis generation grounded in papers 🔹 Scideator: a human–AI interface for using literature as a source of inspiration 🔹 CHIMERA: a dataset for training and evaluating models that recombine concepts to spark new scientific directions The session will also explore how we assess novelty, refine ideas, and design AI systems that truly support scientific creativity. Register to join us! #AIforScience

Exciting progress from one of our community members 🚀 A big step toward AI-assisted scientific discovery! 🤝 Join the AI Physicist Community: https://lnkd.in/evgDC9-s

Can we build a model that predicts physics the way LLMs understand language? That’s the vision behind the General Physics Transformer (GPhyT), and the topic of our next FirstPrinciples Talk. On Tuesday, November 18 at 10 am (ET), we welcome Florian Wiesner, ML Research Scholar at the University of Virginia , to share the first steps toward a physics foundation model. Register to join us! Read "Towards a Physics Foundation Model" ahead of the talk 🔗 https://lnkd.in/eDQKBrrw #AIinScience #PhysicsAI #FoundationModels #ScientificML #ComputationalPhysics #FirstPrinciplesTalks #Webinar

We’re thrilled to welcome Catherine Patterson to the FirstPrinciples team as our new Development Manager 🚀 Our team is growing! Explore careers at FirstPrinciples 🔗 https://lnkd.in/ekZB3bts #TeamAnnouncement #AIinScience #Research #Hiring

How can physics guide the next generation of AI? 🤔 Join us this Thursday, November 13, from 1-2 PM ET for our next FirstPrinciples Talk with Tess Smidt, Associate Professor at the Massachusetts Institute of Technology and Principal Investigator of the Atomic Architects group. In her talk, “Context, Culture, and Craft in Effective AI for Physics,” Tess will explore how embedding physical structure into AI systems (particularly through Euclidean neural networks) can make models more interpretable, reliable, and aligned with the symmetries that govern the natural world. Register to attend! #Webinar #AI #Physics #AIforScience

Exciting progress at the intersection of AI and scientific research 👇 Boris joined us for a FirstPrinciples Talk earlier this year to discuss Denario’s potential as a multi-agent research assistant and what it means for the future of AI in science. Congratulations to the Denario team! #AIinScience #OpenScience #Agents