Social media users raised concerns about the AI tool’s generation of images depicting historical figures, such as the U.S. Founding Fathers, as people of color, labeling these representations as inaccurate.
In a statement posted on X on Wednesday, Google acknowledged the issue, stating that while the AI feature can generate a diverse range of people, it was “missing the mark” in historical depictions. The tech giant assured users that it is actively working to improve these inaccuracies.
An updated statement from Google on Thursday confirmed the pause of Gemini’s feature for generating images of people. The company pledged to release an “improved” version soon.
Gemini, formerly known as Bard, launched its image generator tool at the beginning of February. However, it faced immediate challenges amid criticism over historical representation accuracy, coinciding with Google’s efforts to compete with Microsoft-backed OpenAI.
During testing on Thursday morning, Gemini failed to generate any images at all, according to a CNBC reporter.
Meanwhile, OpenAI introduced Sora, its new generative AI model capable of producing video content from users’ text prompts, further intensifying competition in the AI landscape.
Jack Krawczyk, senior director of product for Gemini at Google, reiterated the company’s commitment to addressing representation and bias seriously. He emphasized Google’s dedication to refining image generation capabilities, particularly in historical contexts that require nuanced understanding.
Google’s rollout of Gemini, its most advanced AI model, occurred at the close of 2023. Additionally, this month saw the rebranding of Bard, Google’s ChatGPT competitor, to Gemini, alongside the launch of a subscription service for a more powerful version of the AI model.