Google has upgraded its AI-powered image-generation tools, months after suspending use of the technology due to issues with its depictions of people.
In February, Google apologized and suspended the creation of AI-generated images showing people in its Gemini chatbot after a number of “inaccurate or even offensive” results were shared online.
Today, the tech giant relaunched the tool with a new version of its image-generating technology, called Imagen 3, which includes new security features, and with access to creating images of people now limited to paid Gemini subscribers.
Google said its new image generation model will have “built-in safeguards and adhere to our product design principles.”
“We have made significant progress in providing a better user experience when generating images of people,” the tech giant said in a blog post.
“We do not support the creation of photorealistic, identifiable individuals, depictions of minors, or excessively gory, violent or sexual scenes.
“Of course, not every image created by Gemini will be perfect, but we will continue to listen to feedback from Gemini Advanced Early Access users as we continue to improve.
“We will gradually roll out this feature, with the aim of offering it to more users and languages soon.”
The tech giant confirmed that the ability to generate images of people would only be available in English for Gemini Advanced subscribers, but the generation of images that do not involve depictions of people would be available to all users.
Depictions of people in AI-generated images in Gemini were disrupted in February, when Gemini users began reporting that the chatbot was generating images showing a range of ethnicities and genders, even when it was historically inaccurate — for example, prompts to generate images of certain historical figures, such as the Founding Fathers of the United States, returned images depicting women and people of color.
At the time, some critics accused Google of anti-white bias, while others suggested the company appeared to have overcorrected due to concerns about long-standing racial bias issues within AI technology, which had previously seen facial recognition software struggle to recognize, or mislabel, Black faces, and speech recognition services fail to understand accented English.
At the time, Google apologized — CEO Sundar Pichai said the incident was “unacceptable” and that the company had “made a mistake” — and pledged to fix the problem.
Alongside the image generator update, Google also confirmed that it’s starting to roll out what it calls Gems, smaller, customizable versions of Gemini that users can customize to become their own personal AI experts on a topic.
The rollout, which will initially be for Gemini Advanced users, will include a set of pre-built Gems to give users an idea of how they can be used, including a writing editor and a coding expert.
