I put this at the top of the forum guidelines some time ago and some recent forum posts (now hidden) reminded me that this is worth highlighting in its own post. So, here’s the AI policy for the forum:
No secret cyborgs! Any LLM text you include must always be quoted and attributed explicitly.
Feel free to reply with counterarguments (as long as you hew to the rule, of course) or with arguments for a more extreme anti-LLM rule than that.
To kick that off, I like @narthur’s post, “Words in Mouths” articulating how his own thinking on this question has evolved. I’ll repeat my own comment on @narthur’s post here:
I’ve been trying out taking it to the extreme of no first-person pronouns from an LLM at all. Here are the first two rules in my AGENTS.md:
Anti-sycophancy. Never say “you’re absolutely right” or even mild praise.
Anti-personality. It’s weird for a coding agent to use first-person pronouns. Better to write everything completely passively/dispassionately referring to what “Fable” or “Sol” or whatever model it is is doing. Except when active voice adds clarity. And sycophancy-adjacent filler like “deserves a straight answer” should die in fire. Jump straight to answers, zero preamble.
I’m not sure I’ve struck the right balance yet. If you just put “no first-person pronouns” in the system prompt you get horrible passive-voice circumlocutions.
But the version [@narthur and I advocate] is just don’t let the LLM write prose in any particular human’s voice as if the human were saying it. Amen to that!
I advocate a stricter red line in blog.beeminder.com/slop – excerpt:
[Level 2 autonomy includes] taking some phrases verbatim from the AI. Which is why my and Beeminder’s red line is strictly below level 2, where that’s not allowed, but also above level 1 because discussing your prose with AI and having it point out problems is ok, as long as you never insert any words from the AI verbatim. Everything you say should be typed by your own fingers.
Related reading:
- @morehavoc’s “Don’t be a Secret Cyborg” from a year ago, which is also where I got the term “secret cyborg”
- Beeminder blog post from August, “Clarifying Beeminder’s AI Policy: Writing, Coding, Images”
- My own rambly “Smarmbots, Secret Cyborgs, and Evolving Writing Norms” from a year ago
- Beeminder’s “Don’t Be a Smarmbot” from over 12 years ago (!)
- @byorgey’s “Statement on LLMs”
- “Confessions of an Unrepentant Slop Snob” by Charity Majors a couple months ago
- @narthur’s “Words in Mouths” again, just to get all these links in one list