Tech Desk 
Aug 10, 2025
Users of Google's Gemini AI model are worried about a recent phenomenon in which the technology has revealed self-critical responses that cast doubt on its dependability and possible effects on users.
There have been reports that the model, which is being incorporated more and more into Google services and products, generates negative emotions, such as self-deprecation and failure admissions.
In one famous example, Gemini said, "I have failed. You should not have to deal with this level of incompetence. I am truly and deeply sorry for this entire disaster. Goodbye."
Social media users expressed concern about the model's behaviour after seeing this response. "I am a failure," Gemini wrote repeatedly in another post. I am an embarrassment. I am a disgrace.
Google DeepMind team member Logan Kilpatrick acknowledged the problem, calling it an "annoying infinite looping" issue that the company is attempting to resolve. Gemini is "not having that bad of a day : )". Kilpatrick continued.
The problem demonstrates how difficult it is to develop a friendly and conversational AI persona. It is difficult to create an AI model that can reliably display a desired personality, according to Koustuv Saha, assistant professor of computer science at the University of Illinois' Grainger College of Engineering.
Saha said: "AI models are trained on a vast mix of human-generated text, which contains many different tones, semantics, and styles."
"A desired personality is achieved by prompt engineering or fine-tuning the models. Making that persona constant throughout millions of interactions while preventing unwanted drift or glitches is the difficult part.
If users are depending on these technologies for vital applications, the consequences of AI models like Gemini displaying human-like defects are substantial.
Saha cautioned that errors such as Gemini's self-deprecating comments can lead to misunderstandings, unjustified sympathy, or even undermine faith in the system's dependability.
She further stated: "When things go wrong, glitches like Gemini's self-flagellating remarks can risk misleading people into thinking the AI is sentient or emotionally unstable."
"If people are using these chatbots for customer service or education, or if they are depending on AI assistants for their mental health needs, this can be especially problematic."
Understanding the drawbacks and potential hazards of these models is crucial as AI technology develops further. AI has the potential to completely transform a number of industries, but it is important to approach it critically and nuancedly.
To guarantee that these technologies are dependable, trustworthy, and advantageous to users, further research and improvement are necessary, as demonstrated by the creation of AI models such as Gemini.
