ChatGPT confuses users by speaking 'Spanish' as ​​AI gets out of control

ChatGPT confused users by answering questions in Spanish as highly touted artificial intelligence (AI) tool was hit with embarrassing glitch

Reports have emerged on social media of an OpenAI-owned chatbot speaking gibberish, with the US tech giant admitting users were receiving «unexpected responses».

Among the strange responses were messages written in a mixture of English and Spanish, while others repeated the same word ad infinitum.

One example said: “Let me explain that there are more wonders that your articulation craves!”

Gary Marcus, Artificial Expert intellect and professor emeritus at New York University, said ChatGPT has “gone crazy.”

In his newsletter, Mr. Marcus wrote, “These systems have never been stable.”

< p>“No one could provide security guarantees around them. We are still living in the age of machine learning alchemy.”

On the company forum, one user said: “About three hours ago, all my conversations with GPT4 quickly turned into garbage.”

Several users said ChatGPT would launch as normal before the response became meaningless.

One ChatGPT message said: “Day. Height. Chalk. Spark. Line. Ken. Vir. Vane. Cowshed. Bye. Was. Objection.»

OpenAI acknowledged that some users were having problems, but said: «The issue has been identified and is being resolved.»

The problem has been fixed.

ChatGPT is considered a major advance in the development of artificial intelligence that can provide humane answers to detailed questions.

Companies are turning to the bot to automate tasks, using it to write emails and summarize reports. or answer customer questions.

Tech giants have invested billions of dollars in artificial intelligence companies to develop “large language models” that are trained on huge databases containing billions of articles and thousands of books.

Microsoft has pledged to invest $13 billion in OpenAI alone, while Amazon and Google have each invested $1 billion in rival Anthropic.

Elon Musk also launched a competing venture called xAI.

However, skeptics have long noted that chatbots are prone to giving strange answers, especially because bots can make up facts, a process known as “hallucination.”

Last week, OpenAI launched a new artificial intelligence tool designed to create photorealistic videos based only on text clues.

However, some videos contain strange distortions, such as objects, animals or people appearing out of thin air.

Sam Altman, chief executive of OpenAI, has previously said that AI could pose a real threat to the future of humanity.

At the same time, he is raising billions of dollars. develop even more powerful versions of this technology.

Concern about the unknown risks posed by AI was partly at the heart of the government's AI security summit in November and led to an agreement to create a security institute to monitor the threat.

The government is experimenting with the use of ChatGPT on its website.