OpenAI is testing new technique where AI models explain reasoning to each other

From Business Insider: 2024-07-20 09:13:00

OpenAI is testing a new transparency technique, having AI models explain their reasoning to each other. This move is part of the company’s effort to develop an artificial general intelligence that is safe and beneficial. The initiative follows recent changes and concerns in OpenAI’s safety department due to leadership departures.

The technique involves powerful AI models conversing to enhance transparency and reveal their reasoning processes. OpenAI released several safety techniques recently to mark progress towards their goal of creating a safe artificial general intelligence. Concerns about the company’s safety commitment have been raised by departing researchers and experts.



Read more at Business Insider: OpenAI’s New Safety Hack: AI Models Policing Each Other