I find it comical that they pretend not to know how to fix it, but just in case I’ll make it explicit: Go into the text that you silently append to people’s pr

I find it comical that they pretend not to know how to fix it, but just in case I’ll make it explicit: Go into the text that you silently append to people’s prompts, and remove the EDI material. We all know you didn’t train the model on images of POC Nazis and Vikings, and if you’re that concerned about bias then improve the training data. If there are no images available to train on for what you consider to be a problematic bias then leave it alone. It’s probably pretty close to reality and nobody is going to care except people who will simply find something else to complain about anyway.