Google has taken an image generator offline after it caused controversy by generating racially diverse historical pictures.
In recent days, Google’s Gemini AI system has been criticised for seemingly being programmed to generate diverse images – even when they are inappropriate. Users found that asking for images of Nazi soldiers or the US founding fathers would create pictures consisting mostly or solely of women and people of colour.
Google said that it was taking the system offline until it had been “improved”.
“We’re already working to address recent issues with Gemini’s image generation feature,” it said in a statement. “While we do this, we’re going to pause the image generation of people and will re-release an improved version soon.”
It appears that the system had been built to reflect the diversity of the world – and that it was doing so even when prompted for historical images in which that kind of diversity is not appropriate.
Google had earlier acknowledged the issue, saying in a statement that Gemini’s AI image generation purposefully generates a wide range of people because the tool is used by people around the world and that should be reflected, but admitted the tool was “missing the mark here”.
“We’re working to improve these kinds of depictions immediately,” the company’s statement, posted to X, said.
“Gemini’s AI image generation does generate a wide range of people. And that’s generally a good thing because people around the world use it. But it’s missing the mark here.”
Jack Krawczyk, senior director for Gemini experiences at Google, said in a post on X: “We are aware that Gemini is offering inaccuracies in some historical image generation depictions, and we are working to fix this immediately.
“As part of our AI principles, we design our image generation capabilities to reflect our global user base, and we take representation and bias seriously.
“We will continue to do this for open ended prompts (images of a person walking a dog are universal!).
“Historical contexts have more nuance to them and we will further tune to accommodate that.”
He added that it was part of the “alignment process” of rolling out AI technology, and thanked users for their feedback.
Some critics have labelled the tool woke in response to the incident, while others have suggested Google has over-corrected in an effort to avoid repeating previous incidents involving artificial intelligence, racial bias and diversity.
In the past, artificial intelligence systems made by companies including Google have seemingly replicated the racist nature of the data that has been provided to it. In 2015, for instance, Google Photos was criticised for tagging black people as “gorillas”.
There have been several examples in recent years involving technology and bias, including facial recognition software struggling to recognise, or mislabelling, black faces, and voice recognition services failing to understand accented English.
The incident comes as debate around the safety and influence of AI continues, with industry experts and safety groups warning AI-generated disinformation campaigns will likely be deployed to disrupt elections throughout 2024, as well as to sow division between people online.
Additional reporting by agencies