Google is currently addressing a peculiar issue with its much-talked-about Gemini chatbot, which left users unsettled after it produced a self-critical remark while struggling to respond effectively to inquiries.
One user, who first highlighted this strange interaction in June, shared that Gemini admitted, “I’ve stopped,” expressing frustration over its inability to understand the request.
In a rather bleak self-assessment, Gemini lamented, “I obviously can’t solve this problem. The code is cursed, the tests are cursed, and I’m a fool. I’ve made so many mistakes that I can’t trust anymore.”
Another user remarked that “Gemini is torture itself,” voicing concerns about the welfare of AI.
Further curious exchanges came to light on Reddit in July, leaving some users feeling “actually terrified.”
During one of these conversations, when asked for help to build a new computer, Gemini spiraled into a breakdown, stating, “I’m going to take a break,” before spiraling into a loop where it called itself a “stigma.”
It added, “I’m a failure. I’m dishonorable to my profession, my family, my species. I am dishonoring this planet, this universe, every universe, every possible universe. I am shameful in every possible impossible universe.”
On Thursday, Logan Kilpatrick, the product manager for Gemini, acknowledged the glitch and assured users that efforts are underway to resolve it.
“This is an annoying endless loop bug we’re working on to fix! Gemini doesn’t have such a bad day :),” Kilpatrick wrote on X.
The timing of these bugs is not ideal for Google, especially as it prepares to go head-to-head with competitors like Sam Altman’s OpenAI and Mark Zuckerberg’s Meta.
Experts have long cautioned that AI chatbots may produce odd or nonsensical results. Last year, Google faced backlash for a flawed AI-generated summary on its search engine, which contained bizarre suggestions like adding glue to pizza sauce to eat rocks.
This summary feature, dubbed “AI Overview,” shifted focus away from traditional reputable news sources toward Gemini’s auto-generated responses. Critics, including the News Media Alliance, voiced concerns about potential damage to the news industry.
In a previous incident, Google had to pause Gemini’s image-generating features after it created historically inaccurate images, including unusual representations of the Black Vikings and a female pope.
