Google’s bubbling scandal over racial rewriting of history in its generative artificial intelligence tool, Gemini, has finally reached the C-suite, demanding the personal intervention of Sundar Pichai himself.
In a letter to staff leaked to media outlets Semafor and The Verge, the chief executive said their company had effectively betrayed the trust of its users and its own corporate mission by openly exhibiting favoritism towards ethnic minorities and people of color.
“Some of its responses have offended our users and shown bias—to be clear, that’s completely unacceptable and we got it wrong,” he wrote.
Earlier this month, the internet blew up when images of a black George Washington and racially diverse WWII soldiers fighting in Hitler’s Aryan Nazi army went viral. Equivalent attempts to create images of strictly white individuals were refused by Gemini.
Tesla and SpaceX boss Elon Musk, one of the earliest voices recognizing the potential benefit and harm of AI, argued it only proved conspiracy theorists right that whites are being discriminated against as part of the “Great Replacement”.
Google’s A.I. “Gemini” hates white people pic.twitter.com/eAxkO3D5HZ
— Jesse Watters (@JesseBWatters) February 23, 2024
Pollster Nate Silver argued Gemini serves as a cautionary tale of the risks of ceding a lot of power to the whims of a handful of AI engineers, especially in the case of a behemoth like Google that is worth some $1.7 trillion and has access to a vast library of personal data based on the world’s collective search history.
“It’s increasingly apparent that Gemini is among the most disastrous product rollouts in the history of Silicon Valley and maybe even the recent history of corporate America, at least coming from a company of Google’s prestige,” he wrote on Wednesday.
Google did not immediately respond to Fortune’s request for comment.
Americans didn’t need AI’s help to fight over the divisive issue of identity politics. The subject has already reached the confines of corporate boardrooms where some individuals have sought to make diversity, equity and inclusion (DEI) programs politically toxic. Claudine Gay and Alissa Heinerscheid, formerly of Harvard and Bud Light respectively, became poster women for diversity hiring gone wrong.
Into this social tinderbox AI companies have lobbed a technological Pandora’s Box whose outcomes are based on humans training the software through implicit weights and biases. The debate around misinformation and its more malicious cousin, disinformation, has now shifted from the news media to AI companies.
Google CEO worries trust has been damaged
The stakes have only gotten higher as investors, who smell a veritable gold rush, push companies like Google to commercially deploy their generative AI tools quickly. AI chip stock Nvidia, for example, has outperformed the Nasdaq by a factor of 10 since ChatGPT launched at the end of November 2022. Last week it posted the largest gain in market cap in history, adding the value of a company like Coca-Cola in a single day.
Meanwhile, the head start Google held after acquiring DeepMind a decade ago has seemingly evaporated as OpenAI continues to dazzle with its latest breakthrough, Sora. The text-to-video generative AI tool launched on Feb. 15, the same day as the iteration of Gemini that caused the controversy.
What’s the difference between a conspiracy and the truth?
About 6 months. pic.twitter.com/z7FB01tAu5
— Elon Musk (@elonmusk) February 25, 2024
At the heart of the scandal over Gemini-generated images of black Founding Fathers is its product lead Jack Krawczyk, who critics swiftly alleged has a history of espousing politically charged views.
“This nut is a big part of why Google’s AI is so racist and sexist,” Musk wrote of Krawczyk, who one day earlier took his X account private.
For a platform provider like Google that seeks to stay above the fray, Pichai recognized the risks of being tarred with the same brush as many partisan cable media outlets and daily newspapers.
“Our mission to organize the world’s information and make it universally accessible and useful is sacrosanct,” Pichai wrote. “We’ve always sought to give users helpful, accurate and unbiased information in our products. That’s why people trust them.”
The Gemini incident may have tarnished that image.
