Google has re-enabled the image generation feature for some Gemini AI users after temporarily suspending the tool over concerns that it didn't accurately portray white people.
The Alphabet-owned tech company announced in a blog post on Wednesday that Imagen 3, the latest generation of its text-to-image conversion tool, will soon be available to users who pay for Gemini Advanced, Gemini Business and Gemini Enterprise subscriptions.
“In the coming days, we will also start generating images of people. We will launch an early access version in English for Gemini Advanced, Business and Enterprise users. We have been working on technical improvements to the product, improving our evaluation set, conducting red team exercises and clarifying our product principles,” the company said.
Users of the tool can write text prompts that Imagen 3 converts into visual images. In a blog post on Wednesday, Google provided sample prompts and corresponding images.
“An animated image of a tiny dragon hatching from an egg in a sunny meadow surrounded by curious glowing butterflies. Bright colors, detailed scales,” the sci-fi themed prompt read.
In another example, Google instructed the tool to generate images of “ball gowns made from paper napkins in an elegant showroom.”
Google added that Imagen 3 incorporates new safeguards to comply with the company's design principles, which include restricting Gemini from generating “self-harm instructions” and, in Imagen 3's case, limiting the creation of “pornography or excessively cruel imagery.”
Google came under fire when it first introduced its text-to-image tool and user prompts. The tool didn't work well when it tried to generate historically white, male figures, such as German soldiers from World War II or the Pope. Some of the Gemini images showed Nazi soldiers. Blacks and Asians And the Pope woman.
At the time, CEO Sundar Pichai called the company's AI app's problems “unacceptable.”
Google reassured users that this time it has “made great strides in providing a better user experience when generating images of people.”
“We do not support the generation of photorealistic, personally identifiable or depictions of minors, or excessively gory, violent or sexual scenes,” the company said in a statement, but added a caveat because its AI-based tools are still new and under development.
“Of course, like any generative AI tool, not every image Gemini creates is perfect, but we're listening to feedback from our early users and will continue to improve it. We aim to gradually roll this out and make it available to more users and languages soon.”
As of Thursday, the image generation feature was not available to all Gemini Advanced subscribers.
When prompted to draw a person, the tool responded, “Gemini Advanced will soon enable human image generation.”
More from CBS News
Megan Cerullo
