I opened my email today to find a discussion thread that emerged from one guy putting "what is the future of websites" into Perplexity and copy-pasting the output to a mailing list I'm on. Ironically a lot of people had pretty thoughtful human-written responses to it, but personally I have zero desire to engage in a discussion with someone who sends me nonconsensual slop. This did, however, get me thinking more about the emerging AI etiquette debate. People seem to be converging around a few basic rules, which I think end up being fairly similar in the two fields I care the most about, code and writing.
1) Don't make me spend more effort reading something than you did writing it.
If you're going to respond to a question with "Claude says X" just give me the fucking prompt instead. I am perfectly capable of asking Claude myself, and if I'm asking you, I probably already did! Low-effort slop PRs and similar text artifacts fall under this category too. If you vouch for it and all the tests pass, I might merge it if I trust you, but that requires you to have built that trust in the past, which is a much slower, more difficult process.
Things start to get sketchy when you introduce using AI to edit - I don't mind if you bounce ideas off of your agent or use it to fix grammatical mistakes, but I've seen enough purportedly human-written essays that have been edited into the particular smoothness of Claudelish that I think you ought to be pretty sparing with the edits you accept.
2) Using obviously AI-generated images is a pretty strong indicator that I should not care about whatever you made.
If you couldn't put in the effort to find a stock image, it doesn't speak highly of the effort you put into writing something. I generally find AI-generated images, writing, and music distasteful - these are, at best, throwaway artifacts used for a quick meme or prototype, not an indicator of any form of artistry. This kind of falls into the same category as 1: if you didn't care enough to draw it, or sing it, or whatever it, why should I care enough to consume it?
3) Disclose AI use.
Unfortunately some communities (cough cough bluesky) have effectively discouraged this by dogpiling anything that AI has touched, but I'd much rather people be upfront about how much of something they've entrusted to the robot. This allows people to choose to ignore or engage with it on their own terms, and (hopefully) forces you as a creator to think about where your own lines are.
4) Effort has to be visible somewhere.
Lots of things that used to indicate you cared enough to put real work into something don't necessarily work that way anymore. So you should at least care enough that I can't immediately tell you had the robot oneshot your project while you were scrolling Reels.
This all maybe comes back to my previous post about craft, but after being exposed to ten million forgettable vibecoded projects I feel pretty strongly about it. I don't care if you used AI in the process of building your website any more than I care if my handyman used a power drill instead of a hand drill, but if you want me to care about what you're making as "art" or "a product worth my time" rather than the code equivalent of Shein t-shirt, you better take the time you saved on writing code and put that effort in somewhere else. Give me a cool design, show me you thought more deeply about the user, do something fucking interesting.
When it comes to writing, the artifact itself contains the effort, so just write it yourself, please! I want to know what a real, flawed human being thinks about something, in their own words, not extruded through the "make me sound smart" machine.
I won't shame-link here, but you know who you are.
The irony, of course, is that while I was doing a bit of background research (OK, asking Claude to run a search and put the links into a note for me to review, I guess I should follow my own rules) on this, half the shit I found was just AI-generated recaps of this debate. You'd think people trying to write about AI etiquette would do their best to follow it, but once again my expectations appear to be too high. That said, shoutout to Alex Martsinovich who both appears to have written their post on the matter themselves and beat me to the punch on using a Blindsight quote.
I am mildly hopeful that some mix of pressure to actually watermark AI output and the gradual development of herd immunity to slop will result in these rules being more widely followed in the future. It all really boils down to "be respectful of my time and attention", which doesn't seem like too much to ask. I guess we'll see!