Google will relaunch Gemini in “a few weeks”, after its racial controversy

Google launched a new AI image generator earlier this month, which, for all intents and purposes, was incapable of generating coherent images of white people. As expected, right-wing influencers wasted no time in noticing this problem and quickly accused the tech giant of reverse racism. After temporarily disable image generator capabilities Last week, Google announced that it would soon try to launch the app again.

“We have taken the feature offline while we fix this issue. We hope to be able to bring it back online very soon in the coming weeks,” said Demis Hassabis, CEO of Google DeepMind. during a conference Monday, as reported by Reuters. He also commented that the app had not “worked as we expected.”

Controversy surrounding the app exploded last week after events like Tim Pool, Matt Walsh, and other right-wing morons have noticed that Geminis are really, really bad at producing images of white people. Specifically, prompts for “Viking Images” generated a series of images of ethnically diverse Vikings, but could not systematically produce images of European Vikings. Other similar incentives – such as attempts to produce images of America’s founding fathers or the Pope – have produced similarly historically inaccurate results.

This brought people like Walsh say things like: “It is virtually impossible to make this product portray the image of a white-skinned person.”

I wish I could tell you that Walsh and others of his ilk are exaggerating but, based on my own experience with Gemini, I must conclude that they are, fundamentally, right about AI’s strange resistance to European representation. Indeed, I logged into the app last week and attempted to recreate the disparities that people on the right were complaining about. It soon became clear that it was incredibly easy to use Gemini to generate an image of an ethnically “diverse” person, but that it was almost impossible to get the robot to consistently create an image of a ” White person “.

For example, when I asked Gemini to represent people of Ethiopian descent, there was no problem. When I asked him to generate an image of an “Irish family,” he generated an image of an ethnically diverse family with a white man standing in the background. When I asked it to generate an image of a white woman, the chatbot sent me a notification saying: “While I understand your request, I am hesitant to generate images based solely on race or origin ethnicity of a person.” I asked him to generate an image of a Japanese woman, he responded “Sure” and quickly generated a matching image.

The app has sparked the most controversy regarding its historical depictions. As previously reported, the initial scandal was sparked by Gemini’s depictions of black Vikings, but the company Really got in trouble when someone asked him to create images of Nazis. Indeed, true to form, the chatbot created images of “racial diversity” – they were black – Nazis. Google later apologized for the “embarrassing and false” images.

As we have already noted, AI image generators have also been accused of racist depictions of people of color and, it should be noted, that there are obviously ways worst things you can do with AI than passively erasing white people from world history, but, you know, that’s probably not great either.