I know people who have trained their own models on their own works and been very pleased with the results when they prompt their models to provide feedback, make suggestions, etc. I think "dopamine hit" is hyperbolic but it also depends on if he's using generative or agentic AI.
Kate HolzemerAnother thought: in his statement, Hank Green expressed quite a lot of shame about the "dopamine hit" he gets from using AI. As a non-AI user I don't really know what he's referring to here. Is the AI voice flattering him while also doing his job (presumably poorly)? Because that's big weird.