"ChatGPT's Unpredictable Output Spooks Users"

OpenAI's ChatGPT chatbot started spouting nonsensical and surreal responses, prompting investigation and concern among users. Some responses were incoherent, while others were in a different language or simply meaningless. Cognitive scientist Gary Marcus described the behavior as "a warning," highlighting the lack of stability and safety guarantees in generative AI systems. OpenAI acknowledged the issue and is working to remediate it, with speculation that misconfigured settings may have caused the problem. The company has not provided a detailed explanation yet.