My prompt is meaningless. When I tell AI to proofread my writing or to generate a boilerplate email, pretty much no part of my prompt makes it into the final pr

My prompt is meaningless. When I tell AI to proofread my writing or to generate a boilerplate email, pretty much no part of my prompt makes it into the final product. It's a seed of entropy that gets a literary flywheel spinning, but I'm not so pretentious as to assume my 3-word query is somehow unique or even influences the tone of the output. It's not "my" response just because I gave it a question. Pretty much the past 5 years of model finetuning has been predicated on the idea that AI cannot reliably steer itself in the right direction. Models perform better in almost every way when you seed them with additional, non-unique and human-generated training data.