Google Suspends Gemini AI's Image Generation of People Following Historical Inaccuracies

Google has announced the temporary suspension of its Gemini generative AI suite's ability to create images of people. This decision comes as the company works on enhancements aimed at improving the historical accuracy of human depictions generated by the technology.

In a statement shared on the social media platform X, Google described this move as a necessary "pause" to address "recent issues" related to inaccuracies in its image outputs. “While we implement these improvements, we will halt the image generation of people and look forward to reintroducing an upgraded version soon,” the company stated.

Launched earlier this month, Gemini's image generation tool recently faced scrutiny as users shared examples on social media that highlighted its failures—such as depicting U.S. Founding Fathers as American Indian, Black, or Asian. This led to criticism and mockery online, with Paris-based venture capitalist Michael Jackson labeling Google’s AI efforts as “a nonsensical DEI parody,” referring to the concepts of Diversity, Equity, and Inclusion.

On the same platform, Google acknowledged being "aware" that its AI was creating "inaccuracies in some historical image generation depictions," stating: “We are making immediate improvements. Gemini’s AI image generation showcases a diverse range of individuals, which is beneficial as it serves a global audience. However, in this instance, it missed the mark.”

Generative AI models create outputs based on their training data and specific parameters, including model weights. However, these tools have often faced criticism for producing biased outputs, such as sexualized images of women or depicting high-status roles predominantly with white men.

Back in 2015, another AI image classification tool from Google sparked outrage when it mistakenly labeled Black men as gorillas. Although the company pledged to resolve this issue, Wired reported later that the solution was merely a workaround—blocking the technology from recognizing gorillas altogether.