sloptells

AI-written text has tells. They change as models change — “delve” died in 2024, and the list below is what's measurably alive today. We measure them: generate text from current and older models on prompts matched to pre-AI human writing, find what models overuse, check every candidate against acclaimed human prose so we don't flag good writers, and date everything.

This is a style linter, not an AI detector — these patterns read as AI whoever wrote them. If you write with an agent, point it at the api or llms.txt and let it dodge the list while drafting.

check your text

active — strongly overrepresented in current model output — worth avoiding

“rather than” lexical · ×5.0 vs humans
“not just X” lexical · ×10 vs humans · ◑ appears in good human writing too
“That said, …” lexical · ×19 vs humans
“(this) feels like” construct · ×12 vs humans
“the real question/problem is…” construct · ×29 vs humans
“one of those X” construct · ×13 vs humans
“the actual X” lexical · ×8.2 vs humans
“genuinely” lexical · ×51 vs humans
“worth noting/mentioning” construct · ×10 vs humans
“curious what others think” discourse · ×10 vs humans
“in practice” lexical · ×4.0 vs humans · ◑ appears in good human writing too
hedge-adverb pileup lexical · ×2.3 vs humans · ◑ appears in good human writing too
“especially” lexical · ×2.3 vs humans · ◑ appears in good human writing too
“a few things/considerations” lexical · ×2.0 vs humans · ◑ appears in good human writing too
“culinary” lexical · ×14 vs humans
therapist-mode vocabulary discourse · ×11 vs humans
“what (actually) matters” construct · ×13 vs humans · ◑ appears in good human writing too
bold everywhere formatting · only in AI
bullets for a conversation formatting · ×13 vs humans
sentences that march in formation cadence
suspiciously clean punctuation formatting
emoji as structure formatting · only in AI
headers on a social post formatting · only in AI
“the privilege of” lexical · only in AI
“X taught me Y” construct · only in AI
“here's what actually happened” construct · ×17 vs humans
“The lesson? …” construct
“Not because X. Because Y.” construct
“showing up” lexical · ×6.2 vs humans · ◑ appears in good human writing too
the mic-drop closer cadence

saturated — everyone knows, models (or their users) still do it

“thrilled/excited/humbled to announce” discourse · only in AI
journeys and chapters discourse · only in AI
“And honestly? …” construct · ×0.7 vs humans
“here's the kicker/thing” construct · only in AI
“You're not alone/imagining it” construct · only in AI
“no fluff” lexical · only in AI
signposting overuse discourse · ×0.3 vs humans · ⚠ also common in good human writing
“it's not X, it's Y” construct · ×5.0 vs humans

fading — being trained away — declining across model generations

“Great question!” discourse · only in AI
the 2024 hedging cluster lexical

stale — widely believed, no longer true — we publish the correction

typographic quotes formatting · ×15 vs humans · ⚠ also common in good human writing
the em dash formatting · ×20 vs humans · ⚠ also common in good human writing

retired — historical — the tells of model generations past

staccato mic-drop sentences cadence
“delve” lexical · ◑ appears in good human writing too
“tapestry” lexical
“a testament to” construct