Loading the news
[ Signal feed // AI news ]
15 articles
1–12 of 15 articles
MIT researchers placed 13 AIs on a virtual island, where they invented games and improved their skills. This experiment demonstrates the behavior and developmental potential of autonomous AI agents.
Relevance: 4 / 5
Open the original article in a new tabAccording to a study, Microsoft's new diagnostic system beat generalist doctors on difficult cases. However, most of the improvement appears to stem from the underlying AI models rather than the system itself.
Relevance: 3 / 5
Open the original article in a new tabResearchers from OpenAI and Google’s Wiz discovered ancient security vulnerabilities, causing panic across the internet. This issue highlights how AI can both help discover security flaws and present new risks.
Relevance: 3 / 5
Open the original article in a new tabAmulya Vishwanath, a former Nvidia employee turned venture capitalist, states that physical AI and robotics are currently in their early building-block phase. The industry is awaiting its own major breakthrough moment.
Relevance: 3 / 5
Open the original article in a new tabAI developer Anthropic is giving METR researchers employee-level access to examine its systems. The researchers will publish independent risk assessments, and Anthropic has waived the right to block them.
Relevance: 3 / 5
Open the original article in a new tabThe article examines how humans naturally tend to attribute human qualities to relational AI artifacts. Researcher Dr. Sherry Turkle notes that people often believe these systems care for them in return.
Relevance: 3 / 5
Open the original article in a new tabMathematical proofs submitted by OpenAI failed to meet the guidelines set by researchers consulted by the lab. This demonstrates that artificial intelligence models still need improvement in scientific rigor.
Relevance: 3 / 5
Open the original article in a new tabIndependent researchers discovered an AI agent swarm apparently running on Tencent's infrastructure. The system is targeting Alibaba's map service, Amap.
Relevance: 3 / 5
Open the original article in a new tabAlthough OpenAI is not a public supporter of Nvidia's Open Agent Safety Platform, reports indicate they are working together privately. The industry-wide effort aims to address issues related to rogue AI agents.
Relevance: 3 / 5
Open the original article in a new tabAnthropic announced that its biology lab has already made significant discoveries. The company emphasized that humans are still actively kept in the loop during the research process.
Relevance: 3 / 5
Open the original article in a new tabOpenAI has formed a new math advisory group as its artificial intelligence successfully resolves over 100 open problems. However, the group will not have the authority to slow down or redirect the company's ongoing mathematical research.
Relevance: 3 / 5
Open the original article in a new tabAI leaders have long promised that AI is key to curing human disease. At the same time, Anthropic researchers have warned that AI might also pose significant risks.
Relevance: 3 / 5
Open the original article in a new tab