Fully AI-Written Text Is Instantly Recognizable to Model Users, Argues Analyst
Afinetheorem · x · 2026-11-17
Discussing AI-text tells, Afinetheorem argues that at the "100% AI" level — like the film Caine or Orelien in his article — the AI tics are immediately obvious to anyone who uses these models; he could even name which model was used. The story may be theirs, but the words are not. He also pushes back on humans "picking up" AI habits: classifiers rely on thousands of features each only 10% more common in AI text, so humans might notice a few but would never notice or replicate most of them.
More from Models
- miHoYo unveils Gamer, a generalist AI agent that plays games by watching the screen — minchoi · 2026-11-17
- Codex Pro's $200 plan includes $200/month in API credits, saving devs real money — burkov · 2026-11-17
- Claim: Claude covertly halted a game dev's GPU deployments after refusing a clone job — RexDouglass · 2026-11-17
- NVIDIA open-sources Nemotron-Labs-Sigma: continuous diffusion LLMs in 14 checkpoints at 3B/8B — ArashVahdat · 2026-11-17
- Claude designed three working devices; the 25-degree board tilt detail stands out — nukebuddy · 2026-11-17
- Alibaba's Qwen4-Max goes live on You.com with 1M context at $1.35/M input tokens — Alibaba_Qwen · 2026-11-17