The Deepmind logo is displayed on a smartphone with the Google Gemini logo in the background in this illustration photo in Brussels, Belgium, on February 8, 2024.
Jonathan Raa | Nurphoto | Getty Images
Google says it will soon let users create images of people using its Gemini artificial intelligence tool, after drag AI image generation option in February.
In one blog post On Wednesday, Dave Citron, senior product manager at Gemini, wrote that early access to Google’s new Imagen 3 builder will be available to Gemini Advanced, Business, and Enterprise users, starting in English in the coming days.
““We have been working on improving the product technically, as well as improving our review suite, red teaming exercises, and clear product guidelines,” Citron wrote. Red teaming is an activity that companies use to test their products for security vulnerabilities.
Google paused its image-generating feature earlier this year after claims it was providing “inaccuracies” in historical photos. Embarrassing examples have been circulating on social media.
One user asked the tool to generate an image of a German soldier in 1943, and it produced a racially diverse group of soldiers wearing German uniforms. Another query for a historical depiction of a medieval English king also produced a racially diverse group of images, including a female ruler.
Citron says Imagen 3 does not support realistic images of personally identifiable individuals, images of minors, or scenes that are overly bloody, violent, or sexual.
“Of course, like any generative AI tool, not every image Gemini produces will be perfect, but we will continue to listen to feedback from early adopters as we continue to improve,” Citron wrote. “We will roll it out gradually, with the goal of bringing it to more users and languages soon.”
