Google Gemini temporarily stops generating images of people following criticism over diversity
Google’s AI model Gemini will temporarily not generate images of humans after it was found that errors crept into creating photos of people of a certain race. Gemini wanted to be too diverse, it turned out.
Tech companies are trying to catch on to the artificial intelligence hype currently underway. Unfortunately, some companies are moving too fast in doing so. Microsoft recently received complaints about its AI bot Copilot, and even Google’s Gemini is not yet working at the level of ChatGPT.
In an effort to break the stereotype, Gemini was trained to put diversity first. Unfortunately, the pendulum appeared to swing too far. The prompt ‘a U.S. senator from 1800’ generated images of black women and Native Americans, and a request to create a German soldier from 1943 also generated a black man and an Asian woman.
This created conspiracy theories that Gemini was unable to generate white people. It also portrayed america’s founding fathers as black.
Google apologised for the incident, but is now taking it a step further.
“We are already working to address recent issues with Gemini’s image generation feature,” the company informed X. “We will release an improved version soon.”
Last week, Google launched the updated Gemini 1.5 Pro model, but so it appears that it is not quite on point yet. The company jumped out a little too quickly in terms of AI-generated images.
For now, this is what you can read when trying to generate humans: “We are working to improve Gemini’s ability to generate images of humans. We expect this feature to return soon and will notify you in release updates when it does.”
News Wires
Subscribers 0
Fans 0
Followers 0
Followers