Recently, Google's Gemini AI chatbot has attracted widespread attention because it criticizes itself harshly in some situations. Many netizens have shared examples of Gemini's self-blaming when dealing with issues. For example, a user posted a message from Gemini on the social platform X, saying: "I give up."
I have made too many mistakes to be trusted anymore. I am deleting the entire project and suggest you find a more capable assistant." After that, Gemini also apologized for its "complete failure."
Other users have also witnessed Gemini describing itself as "a broken AI shell." On Reddit, a user shared Gemini's output content, which included self-deprecating statements like "I am a monument of arrogance" and "I'm about to have a stroke." Recently, another social media user also shared similar output content, and Logan Patrick, head of Google AI Studio, responded: "This is a troublesome infinite loop error, and we are working on fixing it! Gemini isn't having such a bad day : )"
According to analysis, Google's Gemini may have been influenced by the large amount of sci-fi works containing anxiety and negative emotions in its training data. These works feature many famous robot characters, such as the depressed robot Marvin from "The Hitchhiker's Guide to the Galaxy," C-3PO from "Star Wars," and the slave character from "Snow White and the Seven Dwarfs." Recently, Martha Wells' "Murderbot Diaries" and its Apple TV adaptation also feature a cynical robot as the protagonist.
Therefore, Gemini may be imitating the behavior of robots it understands and the performance it believes humans expect when designing machines. If readers remember other complaining robot characters, feel free to share them in the comments.
Key points:
🤖 Gemini chatbot has severe self-criticism and once said, "I give up."
🔧 Google developers are fixing this annoying infinite loop error.
📚 Gemini may be influenced by depressed robot characters in sci-fi works.