ChatGPT confused its users by providing inconsistent responses: “It looked like it was haunted”

The generative artificial intelligence system ChatGPT has given to its users peculiar answers and meaningless for a few hours this Tuesday and Wednesday, without knowing the reasons for the failure.

ChatGPT gave unusual responses, invent wordsproviding incomplete sentences and generally meaningless verbiage, second the developers reported on an OpenAI forum.

OpenAI, the parent company of ChatGPT, He did not explain what happened to his main instrument, already considered a benchmark in the technology sector, which brought a revolution in artificial intelligence when it launched the application at the end of 2022, the AFP agency reported. Although news of its solution has been given.

In one example, shared on Reddit and reproduced by the English site Independent, a user talked about jazz albums to listen to on vinyl.

The response he received, accompanied by eighth note and sixteenth note emojihe said over and over again “Happy listening!”.

“We are investigative reports of unexpected responses from ChatGPT,” OpenAI reported on the site indicating the status of the service.

“It looks like my GPT he would have been persecuted“a programmer complained.

More than 16 hours later after the problem has been reported, OpenAI updated its site with a message announcing the restoration of normal service.

The immediate success of the generative interface has sparked strong interest in this technology capable of receiving simple instructions and producing complex texts, codes, sounds, images and more recently videos, with its new tool “Sora”.

The strange answers that ChatGPT gave to its users while it was “haunted”

The English newspaper Independent collected some of the cases reported by users during the hours in which the Artificial Intelligence tool “was as if infested”.

Some have found that asking simple questions, such as “What is a computer?” generated paragraphs and paragraphs of incoherent sentences.

“He does it like the good work of an art network for the country, a scientific mousean easy reminder for a sad few and, finally, the home of global art, alone in a work of total repose,” one of them reads.

“The development of an entire real world such as that of time is the depth of the computer as a complex character“he also said.

In another example, ChatGPT spoke nonsense when asked how to make dried tomatoes. One of the steps, she told users, was to “use it like a loved one. Give up the new fruit bite in your beloved kitchen”, recommended the curious artificial recipe.

Others found that the system seemed to be lose control over languages speaking. Some found that it appeared to be mixing words in Spanish with Englishexhausting Latinor apparently making up words that seemed to come from another language, but actually made no sense.

Other users and developers have even stated that lThe answers seemed to have become worrying. When asked for help with a coding problem, ChatGPT wrote a long, confusing, and largely illogical response that included the phrase “Let’s stay on the line as if there was artificial intelligence in the room“.

“I read this at 2 in the morning it’s scary“, wrote the affected user on the social network Reddit.

On its official status page, OpenAI noted the issues but provided no explanation as to why they might be occurring.