Google Pauses Gemini Image Generator Over Historical Inaccuracy Backlash
Google temporarily halts its Gemini AI tool from creating images of people after the feature produces historically inaccurate depictions. The company admits the diversity-focused generator misses the mark and promises an improved version soon.
Google pauses its Gemini artificial intelligence tool's ability to generate images of people following intense social media criticism over historically inaccurate results. Users blast the tool for replacing White historical figures with people of color, highlighting a significant flaw in the system's approach to representation.
The embarrassing blunder reveals how tech giants struggle to balance diversity in AI without creating absurd historical inaccuracies. While Google attempts to avoid the harmful racial stereotypes often produced by other image generators like Dall-E, this heavy-handed approach instead prevents the AI from accurately depicting specific historical contexts, such as generating a correct image of a pope or a 1943 German soldier.
Google acknowledges that Gemini misses the mark and confirms it is working to fix the issue. The company temporarily suspends the people-generating feature entirely, promising to re-release an improved version that better handles the complex relationship between AI training data, racial bias, and historical accuracy.