So, if you didn’t see today, the internet was buzzing with jokes about Google’s new Gemini AI. See, Gemini could make AI generated images and someone noticed…something was off.
It seemed as if Gemini had some very odd ideas about history. More specifically, it would basically avoid making an image of a white man at all costs (even rejecting exact requests for one) and inserted a lot of diversity into situations where there had been none such as during WW2 Germany.
For once, every sector of the media from right wing to left wing all agreed on how stupid this was and attacked Google, although for different reasons. Conservative outlets for changing history by inserting diversity and left wing outlets for inserting diversity in places that was offensive, as with the Nazi imagery.
It was absolutely hilarious.
Interestingly, someone figured out how to get Gemini to tell on itself. Gemini actually edits prompts before passing them to the image generator adding instructions to add diversity or to create images of certain ethnicities.
That is to say: it was programmed to do this.
Unsurprisingly, Google took down the image generation feature since it was so embarrassing. Google continues its downward slide with its awful search results.
I’m definitely not commenting on this more than it was a great day for memes and I think AIs should be accurate.
Have a nice evening.