SAN FRANCISCO: ChatGPT spewed nonsensical answers to users’ queries for hours Tuesday into Wednesday before eventually returning to its senses.
OpenAI, which makes the world-leading generative artificial intelligence (AI) tool, said a software tweak had “introduced a bug with how the model processes language”.
“Upon identifying the cause of this incident, we rolled out a fix and confirmed that the incident was resolved,” it added.
ChatGPT was giving “peculiar” responses, generating non-existent words, incomplete sentences and general gobbledygook, developers using the tool said in a discussion forum on the OpenAI website.
“It gives me meaningless words followed by a bizarre list,” one developer lamented.