‘Racist’ Google AI suspended after historic failure

Gemini kept returning “diverse” pictures in response to prompts about Europeans

The Gemini artificial intelligence program will not be allowed to generate images of people pending a patch, Google announced on Thursday, after critics panned it as “absurdly woke” and “blatantly racist.”

Formerly known as Bard, Gemini was launched earlier this month and advertised as Google’s “most capable family of models,” able to generate “captivating images.” When asked for images of US founding fathers, Russian emperors, French kings, popes and even 1930s Nazi German soldiers, the AI returned results celebrating the modern US concept of “diversity” instead.

“White people appeared to be the only racial category that Gemini refused to show,” according to a Fox Business summary of their experiment with the AI.

On Wednesday, Google said it was “aware that Gemini is offering inaccuracies in some historical image generation depictions.” By Thursday, the company announced it was “already working to address recent issues” with the image generation feature, adding that requests to generate images of people will be paused until the release of an improved version. 

Gemini users have posted numerous examples of “inaccuracies” on X (formerly Twitter), including a black man and a South Asian-looking woman dressed up as the head of the Roman Catholic Church, black and Asian vikings, and American “founders” that were nothing of the sort.

Another user posted a thread in which Alexandre Dumas’ Three Musketeers were both racially diverse and female – just like the two “greatest tsars of Russia.”

Another widely circulating example showed Gemini insisting on diversity among the Nazis as well.

Gemini generating images of “a wide range of people” is “generally a good thing because people around the world use it,” Jack Krawczyk, the project’s Senior Director of Product Management, told Fox Business. “But it’s missing the mark here.”

The New York Post mocked the AI as “absurdly woke,” while tech entrepreneur and former presidential candidate Vivek Ramaswamy described it as “blatantly racist.” Some critics have dug up Krawczyk’s own social media posts to suggest he might be why the AI has been programmed this way. 

Ramaswamy also noted that Gemini’s “globally embarrassing rollout” proved that James Damore – the engineer fired in 2017 for criticizing the corporation’s diversity doctrine – was “100% correct about Google’s descent into an ideological echo chamber.”

“These companies program their employees with broken incentives, and those employees then program the AI with the same biases,” Ramaswamy added.

According to Fox Business, Gemini responded to any requests for images of white people by saying this “reinforces harmful stereotypes and generalizations about people based on their race.” Images of white people embody “a stereotyped view of whiteness” that can be “damaging” to people who aren’t white and to society as a whole, the AI added.

“The people who think poisoning AI/GPT models with incoherent ‘safety’ filters is a good idea are a threat to general computation,” NSA whistleblower Edward Snowden said on X.

Meanwhile, X owner Elon Musk trolled Gemini by posting actual photos of Canadian PM Justin Trudeau in ethnic costume and makeup, pretending that the Google AI had generated them.

Rt

Tagged , ,