Google CEO Sundar Pichai criticized the widely criticized Gemini AI chatbot’s habit of generating “woke” versions of historical figures in a scathing letter to employees. In an email, he called it “totally unacceptable.”
Pichai said Google’s AI team is “working around the clock” to fix Gemini. Gemini was disabled last week after its “absurdly woke” chatbot sparked a social media firestorm with bizarre revisionist photos of black Vikings, a female pope, and even a “diverse” photo German soldiers of the era.
“We know that some of our responses are offensive to our users and show bias,” Pichai wrote in the memo, which was first obtained on Tuesday. “Frankly, that is completely unacceptable. Our judgment was wrong.” semaphores.
The letter marks the first time Pichai has spoken publicly about the PR nightmare that wiped out more than $70 billion in market value in a single day earlier this week, and shows that Google employees are not willing to politicize the technology. This has raised new concerns among critics who say it injects bias.
Pichai added that Google is “already seeing significant improvements across a wide range of prompts.”
“There is no such thing as perfect AI, and we know the hurdles are high for us, especially in this new phase of industry development, and we will keep trying no matter how long it takes,” Pichai said. . “And we’re going to look at what happened and fix it at scale.”
Gemini’s ‘woke’ image disaster undermines public trust in Google’s AI tools at a pivotal time for the company, which just days ago rebranded its chatbot and announced an image generation tool. It hurt even more.
Concerns about potential bias in Gemini’s model training were further exacerbated when social media users resurfaced a series of political tweets by Google Gemini product lead Jack Krawczyk. In an older post, Krawczyk allegedly said that “white privilege is very real” and that America is rife with “terrible racism.”
Scrutiny was not limited to Gemini’s image output. The chatbot sparked further outrage with its text response after refusing to condemn pedophilia and claiming there was “no right or wrong answer” to a question about who was worse, Adolf Hitler or Elon Musk. caused it.
Google’s chief said the company would take a “clear set of actions” in response to the crisis, including unspecified “structural changes, updated product guidelines, improvements to the launch process, and a robust evaluation.” and red teaming, technical recommendations.”
One of Google’s top AI executives said at an event earlier this week that the issue is expected to take “a few weeks” to resolve.
