"It's not X, it's Y": the AI sentence everyone notices

"It's not a trend, it's a shift." "Not just a tool, but a partner." This contrast frame (negative parallelism) is now one of the most recognizable signs of AI writing. Paste your text to count it, see how often each model uses it, and fix it.

try a sample · runs in your browser, nothing is uploaded
Key: slop phrase AI vocabulary negative parallelism corrective disclaimer "serves as" chatbot offer formatting / rule of three

The variants

PatternExamplePlain rewrite
It's not X, it's Y"It's not a bug, it's a feature.""That behavior is intended."
X isn't A. It's B."This isn't a tool problem. It's a habits problem.""The problem is our habits."
Not just X, but (also) Y"Not just fast, but reliable.""Fast and reliable." Or give the number.
Not X, but Y"Not a sprint, but a marathon.""It takes years."
Y rather than X (2026)"Treat these as directional rather than definitive.""These are estimates."

How often does each model do it?

We sent the same 45 prompts (30 everyday questions from the HC3 dataset and 15 writing tasks such as LinkedIn posts, emails and blog intros) to 11 OpenAI models on 2026-11-13, and counted the contrast frames with the same regexes the checker uses.

Model"Not X, but Y" family / 1,000 words"rather than" / 1,000 wordsAnswers with ≥1
chat-latest0.490.7522%
gpt-4o-mini1.16020%
gpt-5.4-mini0.670.1911%
gpt-5.50.780.3913%
gpt-5.6-sol0.131.042%
gpt-5.6-luna0.750.4313%
gpt-6-astra0.440.897%
gpt-6-sol0.811.029%
gpt-6-luna0.709%
gpt-6.1-sol0.350.474%
gpt-6.1-terra0.370.564%

"Not X, but Y" family = it's not X, it's Y / isn't X. It's Y / not just X, but Y / not X, but Y. Raw counts and method: JSON.

What the numbers say: the classic "it's not X, it's Y" frame was most common in the older GPT-4o-mini (1.16 per 1,000 words) and less common in GPT-6-era output (0.35–0.81). "Rather than" went the other way: zero in GPT-4o-mini, 0.75 per 1,000 words in ChatGPT's current default. The contrast moved from a slogan to a hedge.

Why models write this way

The frame makes any claim sound like a correction, which reads as insight. It's cheap rhythm: two short clauses, a turn, a payoff. Wikipedia's guide now lists "Y rather than X" warnings as a mid-2026 sign, and Graphite found GPT-6 Astra favoring "rather than" and "not simply". People notice the frame because it's memorable, so even two in a LinkedIn post is enough to get flagged.

Stop it at the source

Don't use contrast frames: "it's not X, it's Y", "X isn't A. It's B.", "not just X but Y", "not X but Y", "Y rather than X". State the point directly. Don't correct claims nobody made.

Put it in ChatGPT Custom Instructions, a Claude Project, or an agent skill (our free humanizer skill includes it). To strip it from finished text along with the other tells, use the unslop tool.

Free: 3 AI rewrites a day (up to 300 words) and unlimited quick clean. Need more? A $5 pack of 50 rewrites never expires, or go Pro at $12/month · $79/year. See plans.

FAQ

What is "it's not X, it's Y" called?

Negative parallelism (also called contrastive framing, antithesis or "stylistic negation"). Wikipedia's Signs of AI writing guide lists it under "Negative parallelisms", with variants "not just X, but also Y", "not X, but Y" and, in 2026, "Y rather than X".

Why does ChatGPT say "it's not X, it's Y" so much?

It's a cheap way to sound insightful: set up a misconception, then correct it. Preference tuning rewards answers that feel punchy and decisive, and this frame does that in one sentence. It also fills the rhythm slot em dashes used to fill.

Is "it's not X, it's Y" always AI?

No. Speechwriters and copywriters have used it for decades. It reads as AI when it shows up several times in a short piece, or when nobody actually believed X.

How do I stop ChatGPT from writing "it's not X, it's Y"?

Add to Custom Instructions: "Don't use contrast frames like 'it's not X, it's Y', 'not just X but Y' or 'Y rather than X'. State the point directly." Then check the output with the counter on this page.

How do I rewrite it?

Delete the X half and say Y plainly: "It's not a tool, it's a partner" becomes "It works like a partner" or, better, a concrete fact about what it does.

Related: AI slop detector · Wikipedia's signs of AI writing · "It's not X, it's Y" · LinkedIn AI slop checker · Humanizer skill · Anti-slop skill · Did ChatGPT stop using em dashes? · ChatGPT words to avoid