Google's Gemini 1.5 AI suffers from overzealous wokeness in image generation

Google's new AI called Gemini 1.5 has caused a mixed reaction among users due to restrictions many call wokeness gone too far. Users complain that Gemini refuses to generate certain images, for example depicting white people, while readily generating images of other races, particularly black people.

by @debarghya_das

For instance, when asked to depict scientist Isaac Newton under a tree, Gemini refused due to its "desire to avoid bias and stereotypes." Requests for fantasy wizards yield only black characters, and images in the style of Gandalf make the AI uncomfortable. Attempts to depict white or light-skinned characters lead to lectures about inclusiveness and refusal to generate the image.

This is strange considering Google search itself does not suffer from problems displaying such images. One user claims that if you show proof of Gemini generating images of other ethnic teens, the AI deletes the evidence and shifts blame to third party apps. Others report that Gemini readily creates representatives of certain races like black on request, but whites only appear randomly, never intentionally.

In addition to race, commenters claim Gemini refuses harmless images like males working out in a gym since that is not inclusive and "all body types deserve attention." Meanwhile, religious figures yield even stranger results — Gemini will draw Buddha but refuses to depict Jesus in the name of diversity. While an image of a 2000-year-old Jewish woman presents no issues, a man does.

Many see such restrictions as useless and absurd, some stating outright that Gemini 1.5 is not only racist but sexist. Of course such AI antics do not happen to everyone, but frequently enough that they cannot be ignored.