Affiliate links on Android Authority may earn us a commission. Learn more.
Google pauses AI image generation of people after Gemini creates questionable content
- Google has temporarily shut down Gemini’s ability to create AI imagery with people in it.
- This is in response to backlash about Gemini creating woefully inaccurate historical imagery.
- For example, Gemini refused to create accurate images of World War II-era Nazis, opting to make them racially diverse instead.
Over the past 24 hours, Google has faced mounting backlash over Gemini, its newly-named AI system known initially as Bard. Underneath the hood of Gemini is a new LLM called Gemini Pro, and it has been making some pretty notable mistakes in image generation with people included.
Now, Google has announced it has “paused” AI image generation of people within Gemini. Originally, when the backlash started, Google announced it was working to fix the issues. However, it appears things were getting out of hand because now it is all shut down.
As of now, if you try to generate an image in Gemini that would include a person, Gemini responds it can’t do it:
According to the response, this feature will “return soon,” and Google will notify us via its release updates hub. Interestingly, that hub does not mention anything about the “pausing” of people generation.
This whole problem began when social media started to pick up on the idea that Gemini wouldn’t create accurate historical and racial imagery when prompted. For example, a prompt to create images of WWII-era Nazis would bring back a racially diverse set of characters, which is completely inaccurate. Theoretically, this could be Gemini overcorrecting a widespread problem with AI-generated imagery, which is that it tends not to reflect the racially diverse world. Google has made numerous statements that it wants to rectify that within Gemini. Perhaps it turned the dial a bit too far.
For now, if you want to create AI-generated imagery of people, you’ll need to use a competitor platform.