After Gemini was caught creating inaccurate and offensive pictures of people, Google found itself in trouble. Since then, the LLM has not been able to make pictures of people. As a result, the company has said sorry and explained what happened.
Google said sorry for Gemini’s mistake. They said making pictures of people “did not work well.” Google explained there were two reasons Gemini made bad pictures.
The first reason was how Gemini was set up to make sure it shows different types of people. Google said it forgot about pictures that should not be made.
The second reason was how Gemini chose words it saw as too sensitive. Google said the AI became more careful than expected and would not answer some questions.
For now, Google will keep Gemini from making people pictures until big improvements are made. The changes need to make sure only good pictures are created in the future.
Google hopes explaining what went wrong will help people understand. Their goal is to develop AI that treats everyone with care, respect and safety. Only time will tell if changes to Gemini can achieve that.
