"Unprecedented ChatGPT Malfunction Spooks Users with Bizarre Outputs"

Users of OpenAI's ChatGPT reported unexpected and nonsensical outputs, leading to comparisons of the AI assistant "having a stroke" and "going insane" on Reddit. OpenAI acknowledged the issue and is working on a fix. The incident highlights the challenges of understanding large language models, which function like black boxes. Speculations on the cause range from temperature settings to potential bugs in new versions of the AI model. This echoes past issues with Microsoft Bing Chat and has sparked discussions about the advantages of open-source AI models.