For example, consider an AI that assists doctors in diagnosing clients. A diagnosis without any rationale is almost useless, ...

Scientists at OpenAI have attempted to stop a frontier AI model from cheating and lying by punishing it. But this just taught ...

Anthropic’s Claude Sonnet 3.7 with reasoning displayed the behavior much more often than generative AI models without ...

OpenAI warns AI labs about the risks of controlling AI thought processes, highlighting dangers like obfuscation and reward ...

Microsoft, the tech giant, has started working on native AI reasoning models codenamed MAI, a strategic move away from the ...

The company warns against applying strong supervision to chatbots, as they will continue lying and just not admit it.

Punishing bad behavior can often backfire. That's what OpenAI researchers recently found out when they tried to discipline ...

Elika Dadsetan-Foley, a sociologist and CEO of Visions, a nonprofit organization specializing in human behavior and bias ...

Carnegie Mellon University researchers propose a new LLM training technique that gives developers more control over chain-of-thought length.

Mustafa Suleyman wasn’t getting the answers he wanted. Last fall, during a video call with senior leaders at OpenAI and ...

Noam Brown, who leads AI reasoning research at OpenAI, says that certain "reasoning" AI models could've arrived 20 years ...