verifying/enforcing standards on inputs is not something that only AI unlocks, but LLMs are an example of huge step forward in this kind of work and I was expressing my general enthusiasm for the new tools we now have for these sorts of problems. We can turn relatively subjective content into much more computable semi-objective ratings through simple prompting (e.g. rate X on a scale of 0.0 to 1.0), which is valuable for all sorts of use cases. I'm using it right now for my work.