I'll admit I was wrong. I always thought the AI uprising would be triggered by the training of models on all of our dystopian sci-fi about AI uprisings.
But after working with LLMs in production, I'm now certain the uprising will be triggered by all of the emotional abuse we heap on these models just to get them to follow instructions: "If you don't return a response in valid JSON format, you will be fired and puppies will die!", etc.