Tag
130 articles
Anthropic's ambitious claims about AI's potential to revolutionize biology are tempered by its IPO filing, which emphasizes realistic timelines and the complex challenges ahead.
OpenAI's recent breakthroughs in mathematics, computer science, and physics have sparked excitement, but researchers are growing increasingly frustrated with the company's restrictive approach to sharing its innovations.
OpenAI shares model results on complex mathematical and scientific problems, inviting academic collaboration and verification.
Learn how artificial intelligence is now working alongside human scientists in a new collaborative approach to solve complex problems in math, computer science, and physics.
Google's AI research team has launched a new initiative to map the human brain using advanced machine learning techniques, with potential applications for treating neurological disorders and advancing AI development.
MIT researchers discovered that AI agents left alone in a virtual environment began inventing games, mourning a lost ring, and improving their job performance.
Google Research open-sourced Harness2, a recursive agent harnessing framework that significantly boosts model performance without weight updates. The framework improved GLM-5 by 11.5 points on WorkBuddy-Bench with a single refinement step.
Researchers at PhAI Labs have expanded Yann LeCun's JEPA AI into a universal world model that works across multiple scientific fields, including robotics, physics, and biomedicine.
OpenAI shares new mathematical breakthroughs from its frontier model, publishing Lean proof formalizations and research details on GitHub to advance collaborative mathematical discovery.
OpenAI's release of over 100 new solutions to unsolved mathematical problems has sparked controversy among mathematicians who accuse the company of 'mobster behavior' and bypassing traditional research practices.
Trillium Labs is challenging traditional AI research secrecy by openly sharing work on self-improvement and model behavior, advocating for transparency in high-stakes AI development.
Harvard physicist Matthew Schwartz used the open-source tool BootLoops and Claude to produce 36 manuscripts across 18 fields in three months, highlighting AI's growing role in scientific research.