GenAI has found the most adoption in similar situations: where the downstream effects of the fact that its outputs bear only a passing resemblance to facts doesn't matter.
This makes it excellent at, for example, demos: as long as you don't think about it too hard, the output looks "good enough."
