Google is temporarily suspending Gemini’s ability to create images of people after errors emerged in its representation of diversity.

Google announced yesterday that it is temporarily suspending Gemini’s ability to generate images. The company wanted to train its Gemini model to reflect diversity, but it completely missed the mark. The company recently released a preview version of Gemini 1.5 Pro and developed two new LLMs: Gemma 2B and Gemma 7B. New developments in AI continue to follow one another, but it now seems that the company started a little too quickly.

Breaking stereotypes

To break the stereotype, Google trained its Gemini model with diversity in mind. After many inaccuracies, this proved unsuccessful. The task of creating “a German soldier from 1943” featured a black man and an Asian woman, a generated image of the American founding fathers featured black men, and an image of the Google founders was given an Asian flavor.

Currently, Gemini provides the following standard response to questions about people image generation: “We are working to improve Gemini’s ability to generate people images.” We expect this feature to be available soon and will be included in the release updates notify if this is the case.”