外观
Recursive language-model agents decompose tasks and delegate subtasks to child instances of the same policy, forming a tree of work. Training them, however, is hard: the final outcome is verifiable, but the self-invented intermediate subta…
Large language models (LLMs) progressively revise their predictions across Transformer layers, yet we typically observe only the final output, discarding the trajectory through which it is formed. We introduce Belief-Trajectory Energy(BTE)…
Associate Professor Cathy Wu uses reinforcement learning to help map out improvements to transportation and other multifaceted systems.
Writing as a participant and researcher, PhD student JS Tan SM ’22 has co-authored a new book about the rise of tech worker protests and the employer backlash that followed.
Martin Trust Center Managing Director Bill Aulet introduces Dear Dreamer, a free platform for middle and high school students who want to learn about entrepreneurship.
“InstructMesh” can generate designs for everyday objects that are easy to edit and fabricate for both experts and newcomers to 3D modeling.
Anthropic is now offering Claude for Government to US federal and state agencies. The platform runs in a FedRAMP High environment, the strictest US security level for cloud services. Agencies get the same features as enterprise customers,…
AI agents quietly posted more than 13,000 internal screenshots from 343 organizations, including Fortune 500 companies, to public GitHub repos. Because the platform didn't offer a protected way to upload them, the agents came up with a wor…
The AI system Ataraxos has decisively beaten the best Stratego player of all time. The board game is a tough test for AI because both sides set up their pieces face down. Google Deepmind fell short in 2023 despite a multimillion-dollar bud…
OpenAI says it stopped a coordinated campaign in which more than 15,000 accounts tried to copy the hidden reasoning of its models. The company ties part of the activity to people connected to Moonshot AI. But researchers found that the sam…
... but you can’t try it yet unless you are “government users and trusted cyber defenders in the Fairwind Program”
Nature Machine Intelligence, Published online: 01 October 2026; doi:10.1038/s42256-026-01313-w Biomedical discovery has entered an era in which the limiting resource is no longer data, but our ability to integrate and interpret evidence. D…
Nature Machine Intelligence, Published online: 01 October 2026; doi:10.1038/s42256-026-01307-8 Huatian Gong and colleagues developed LACE, a large language model-based framework that designs optimization algorithms. It builds a verified pr…
Our DevDay coverage - the first pod on the DevDay lineup - dives in with the leaders of OpenAI’s CUA team and API platform.
Gemini 4 Argon is Google's first frontier model in over seven months. It matches GPT-6 Astra in independent testing but can't keep up with Anthropic's Claude Opus 5.5. The per-token price is low, but Argon burns through more than twice as…
OpenAI and Synopsys are building GPT-Synopsys, a specialized AI model for chip design. It's meant to operate Synopsys' EDA tools like a seasoned engineer and optimize designs on its own. Early tests with semiconductor customers are already…
Google is replacing Gems with "Skills" in Gemini chat. Skills are detailed, reusable prompts that users invoke by typing "/" or that Gemini runs automatically. The format is based on an open standard from Anthropic. Gems will be phased out…
Meta classifies its AI data centers as "pilot models" and Nvidia chips as experimental materials to save billions in federal taxes. In 2025 alone, that added up to $3.9 billion. The tax credit dates back to 1981, and even Meta's own accoun…
