“My GPT is haunted.” What strange signals AI users noticed, for several hours

The well-known generative artificial intelligence (AI) interface, ChatGPT, went “crazy” for a few hours, where it started giving nonsensical answers. Users were stunned to receive sentences devoid of any coherence or logic.

ChatGPT went “crazy” for a few hours PHOTO Shutterstock

OpenAI, the Silicon Valley company that developed ChatGPT, was quickly alerted to these unexpected responses and announced that it was investigating the issue. Shortly after reporting the incident, they received assurances that work was being done to resolve the situation, according to AFP.

Numerous users have shared screenshots of examples of ChatGPT's bizarre responses online. This advanced technology enables the generation of incredible quality content, including text, audio and video, on demand.

Incoherent and incomprehensible answers

On the forums for developers using the OpenAI tools, a user described a “strange” behavior of ChatGPT. One user, “IYAnepo”, noticed that ChatGPT was generating nonexistent words and omitting or distorting words, causing confusion.

“One might think that I specified such instructions, but that is not the case. I feel like my GPT is haunted”he wrote.

Another user, “scott.eskridge”, complained that all conversations with ChatGPT “quickly turns to nonsense.”

He copied a fragment of an answer of the interface – “bit dwarves and the list is one of the strangers and the Internet where the currency and the person of the cost is one of the friends and the currency. The next time you look at the system, the exchange and the fact, don't forget to give.”

OpenAI didn't elaborate on the exact nature of the incident, but pointed out that these AI systems, even generative ones, don't have consciousness and don't really understand what they're saying.

Gary Marcus, an AI specialist, believed that this incident should be seen as a “red flag”. He emphasized the need for more transparent and easier-to-maintain and troubleshoot technologies to handle such issues in the future.

“These systems have never been stable. No one has ever been able to put security guarantees on these systems.”he wrote.