Speaking at the DBS Asia Future Forward conference in Singapore on October 12, TikTok CEO Shou Zi Chew said AI will deliver real scientific breakthroughs within the next two to three years, naming physics, biotechnology, chemistry and materials science.
OpenAI has released AI-generated solutions to 377 hard math and physics problems at once, all produced by a single agent from one prompt using a frontier model not yet public, according to Scientific American.
Nvidia is in talks to acquire or substantially increase its stake in the open-source AI lab Reflection AI, which was last valued at about $25 billion, the Financial Times and Reuters reported on October 10.
In a long post on X on Saturday, October 10, Microsoft CEO Satya Nadella called for a fundamental rethink of how AI is deployed, built around a “trust architecture” whose core is an emergency stop switch on every AI model.
On October 10, TestingCatalog noticed that an unannounced model, gpt-rosalind-discovery, had quietly appeared on OpenAI's API pricing page under “Life Sciences Models”, priced identically to gpt-rosalind-research.
An active-learning loop combined with geometric deep learning improves reaction prediction for drug chemistry when data is scarce, as AI learns to judge where a molecule should break.
The "2-step" method makes liquid chromatography retention-time prediction for small molecules work out of the box, with no retraining for new systems, a direct benefit to drug discovery and natural-product research.
Fiderer et al. prove in PRX a fundamental trade-off between an agent's predictive power and its energy efficiency, unifying prediction and forgetting through a thermodynamic objective function.
University of Maryland physicist Matt Landreman used GPT-6 Astra Pro to derive, in 33 minutes, analytic MHD force-balance solutions with asymmetric, three-dimensional nested magnetic surfaces, disproving the 1967 Grad conjecture.
In math manuscripts OpenAI released on October 6, a 31-page paper claims a complete proof of the Cannon conjecture, accompanied by a formal Lean verification.