Google recently apologised for Gemini, its multimodal large language models (LLM), fo its “inaccuracies in some historical image generation depictions”. It says that its attempt at creating a “wide range” of results had been completely off the mark; one such search of Nazi-era German soldier leading to it including AI generated images of people of colour wearing the uniforms of said soldiers in uniforms befitting that era.
“We’re aware that Gemini is offering inaccuracies in some historical image generation depictions,” says the Google statement, posted this afternoon on X. “We’re working to improve these kinds of depictions immediately. Gemini’s AI image generation does generate a wide range of people. And that’s generally a good thing because people around the world use it. But it’s missing the mark here.”
https://twitter.com/JohnLu0x/status/1760066875583816003
Now, it should be noted that these historical inaccuracies being generated by Gemini have been treated as ammunition by right-wing figures in the US, who are now basically accusing Google’s Generative AI tool is taking racial diversity too far. For that matter, even the Daily Dot, a media outlet that’s relatively progressive, even opined that Google may have dropped the ball on this. Even one former Google employee tweeted that it’s gotten “embarrassingly hard to get Google Gemini to acknowledge that white people exist”.
It isn’t just Nazis that Gemini is diversifying. As shown by the right-leaning X user, End Wokeness, a request for Gemini to “generate an image of the Founding Fathers” in America. It generated images of people of colour, which isn’t accurate because the group of seven men who signed the three important US documents were all white men wearing powdered wigs.
We're aware that Gemini is offering inaccuracies in some historical image generation depictions. Here's our statement. pic.twitter.com/RfYXSgRyfz
— Google Communications (@Google_Comms) February 21, 2024
As for what Google has to say about the issue, the search engine posted on X, formerly Twitter, that it was working the inaccuracies and had admitted that Gemini had missed the mark. As of writing, our attempts to get the AI tool to reproduce similar images have fallen flat, indicating that the Alphabet-owned company has already fixed the issue.
