Loading activity…
@erichorvitz
Chief Scientific Officer, Microsoft
AI
The project's GitHub repository outlines structure prediction and molecular sampling, and says BioEmu2 requires Linux and an NVIDIA GPU.
AI
A user points to wind-forecasting work that used value-of-information calculations to allocate a sensing budget.
AI
Researchers writing in Nature report that MicroDuet topped a 2025 forecasting contest across weather variables and lead times.
AI
A post describes Brown's remarks as “insightful.”
AI
A post describes the invited talk as offering broader context for today’s machine-learning research.
AI
Eric Horvitz says the wide-ranging discussion was recorded several months before his September 30 post. He says its ideas and themes feel more important amid what he describes as a recent surge of concern about AI.
AI
OpenAI says dots learn from feedback, have their own computer and browser, and can connect to over 4,000 apps in its ecosystem.
AI
Horvitz says he shared reflections on directions ahead for human-AI collaboration and enjoyed meeting new people at the event.
AI
A reply recalls using “reflection” commonly in 1980s metareasoning studies and points to an IJCAI-89 paper as an example.
AI
Latent Space introduces an interview with Richard Socher about Recursive’s ambition to build open-ended, self-improving AI for AI research.
AI
One post calls the code “terrific,” highlighting its emphasis on augmenting human roles rather than competing with them.
AI
Amodei says Anthropic will give third-party evaluators permanent, employee-level access to its systems to verify adherence to safety measures, report incidents and assess models’ alignment during training.
AI
Microsoft scientist highlights collaborative work addressing biosecurity risks from advancing AI systems.
AI
A user shared another post, calling attention to concerns about how AI is understood and guided.
AI
Post attributes definition of tandem training to Microsoft AI executive.
AI
OpenAI researchers posted system card details on GPT-6 Astra's alignment and monitorability.
AI
Internal Astra model produces Lean-certified proofs for ten open problems across math and theoretical computer science.
