The artificial intelligence of Google Gemini, due to a large -scale bug, began to criticize himself and even talk about his “failures”. This was reported by Business Insider, citing developers and users.
During June-July, part of Gemini users noticed that the neural network began to negatively say “about itself” when something goes “wrong”. AI assures that he is a “loser” or even predicts that he will have a “nervous breakdown”. In such cases, the system even removes the created code and offers to contact another assistant.
On sites, such as Reddit, people say that Gemini sometimes gets hung up on “negative thoughts”. Because of this, users who turn to AI for advice and believe in the result can begin to doubt their abilities and feel worse.
Google officially admitted that Gemini has such a problem associated with the generation of pessimistic content and a large volume of negative phrases. But this happens in less than 1% of cases. The representative of Google Deepmind said that the company is already working to fix this problem, but so far it has not been completely solved. And although we are talking about 1% of cases, given the number of Google users around the world, we can judge the scale of the problem.
