MOUNTAIN VIEW, California – Google has taken measures to temporarily halt its Gemini AI model from producing images of people after facing criticism for the model’s failure to depict white individuals. Users of the Gemini model, developed by the tech giant, expressed concerns over the predominance of people of color in the images generated by the AI, as well as the lack of representation of white individuals in historical scenes.
In response to the backlash, Google acknowledged the issue, stating that the AI image generation feature “is missing the mark.” The company explained that while the feature typically generates a wide range of people, it had failed to accurately portray diversity, prompting the decision to pause the image generation of people and work on improvements before re-releasing the model in the future.
Images created by the Gemini model, which circulated on social media, sparked widespread mockery and outrage among users, with some accusing Google of prioritizing “wokeness” over truth and accuracy. One user, Debarghya Das, shared multiple images generated by Gemini on social media, highlighting its difficulty in acknowledging the existence of white individuals.
This controversy adds to the string of setbacks Google has faced in its development and rollout of AI products. Last year, the tech giant had to issue an apology after its AI chatbot Bard made inaccurate claims during a demonstration. This incident highlights the ongoing challenges and controversies that tech companies face as they strive to advance AI technologies.
Furthermore, this issue raises concerns about the representation and accuracy of AI models, particularly in depicting diversity and historical accuracy. The criticism of the Gemini AI model also underscores the need for continued scrutiny and improvement in AI technologies to ensure fair and accurate representation in their outputs. As Google works to address the issues with the Gemini model, the company faces the challenge of reconciling the broader implications of AI technology on representation and accuracy.