Google’s Powerful AI Image Generator Is Coming to Gemini
Google will once again allow users of its Gemini AI service to generate images of humans after a debacle earlier this year when it was accused of “erasing” white people.
The furore in February prompted Google to disable Gemini’s AI image generator but as of yesterday (Wednesday), users who pay to use the chatbot once again have access to the feature and free users will gain access to it in the coming days.
The foundation model for Gemini’s AI is Google’s Imagen 3 which the tech giant released recently and boasts impressive results.
In PetaPixel’s tests, Imagen 3 does appear to be a quality text-to-image model that rivals Midjourney or OpenAI’s DALL-E. What’s more, Imagen 3 is currently free-to-use unlike the aforementioned.
“With Imagen 3, we’ve made significant progress in providing a better user experience when generating images of people,” says Google senior director Dave Citron.
He adds that users won’t be able to generate photorealistic images of public figures, minors, or any violent imagery.
Google says it has made “technical improvements” to Gemini ahead of the rollout, as well as “improved evaluation sets, red-teaming exercises and clear product principles.”
Google had an embarrassing moment at the first rollout of Gemini’s AI image generator which seemingly overcompensated for racial biases.
For example, a request for imagery of Vikings, the U.S. founding fathers, and catholic popes brought images of women and men of varying races.
It raised concerns over whether Google can maintain its position in the age of AI. Its chatbot, Bard, also stumbled when it gave an incorrect answer about the James Webb Telescope during its initial outing.
Citron admits that Gemini still isn’t perfect and will be listening to user feedback for clues on how to improve it.
Other Features
As well as the AI image generator, Google says that customers can now start creating “Gems”, which are customized versions of Gemini that can be used for specific tasks.
“You can create any Gem you dream up: a gym buddy, sous chef, coding partner, or creative writing guide. They’re easy to set up, too,” Google writes in a blog post.
Image credits: Google.
