Historically inaccurate images and text generated by Google's Gemini AI have "offended our users and shown bias," CEO Sundar Pichai told company employees in an internal memo obtained by The Verge.
Last week, Google stopped Gemini from generating images after it was discovered that the model produced offensive or inaccurate images. Google apologized and said it would re-enable image creation in the coming weeks. The internal memo released on Tuesday shows just how significant the flaw was in a product that is already commercially available.
The CEO tried to emphasize to the company's employees that "the bar is high" meaning of course the competition.
In the note, Pichai says the company is "working around the clock" to address "problematic text and image responses in the Gemini app." It doesn't say that Google has solved the problem. "No Al is perfect, especially at this emerging stage of industry development, but we know the bar is high for us."
Sundar Pichai's memo to Google employees reads:
Hi to all
I want to report the recent issues with problematic text and image responses in the Gemini (formerly Bard) app. I know some of his responses have offended our users and shown bias — to be clear, this is completely unacceptable.
