It would be interesting to track OpenAI outages vs activity on certain topics on Twitter/Reddit to get a vague metric of what % of the comments are AI generated.