Etiquette.

Don't send a full LLM-generated Deep Research article to someone before you've actually read it yourself. Assuming your prompt was good enough that the output must be too, and then gambling 10, 15, or 20 minutes of someone else's time to save yourself the same? That's just rude.

I kind of wish this were self-explanatory, but over the past year I've been sent an "amazing" AI-generated article more than 10 times by people who hadn't actually read the full thing themselves. Not even scanned it. That's a lot of potentially wasted minutes.

I'm not sure where this comes from. My best guess is that people play around with LLMs, figure "I'm so good at prompting," read one or two articles it generated, think "this is so profound," and then just trust whatever it spits out next.

Time is scarce. That's a good thing. But when it takes you a minute to create a prompt and the output becomes 20 minutes of reading, please spend the time making sure it actually says what you thought it would. When in doubt, don't send.