The problem with AI such as GPT4 is training data. As it usage increases, most of its training data will be data generated by GPT4, hence creating a positive feedback loop. This would actually increase the value of human knowledge workers.
Joking, but training data isn't as big of a problem as you suggest. Since so much about AI is still very centralized it's relatively low effort for them to tag generations in a way that makes sense, avoiding a lot of effects of a loop.