How AI detectors like Pangram work: models catch tics even Shakespeare wouldn't notice
Afinetheorem · x · 2026-09-22
Afinetheorem explains how Pangram-style AI text detection works in a 4-part thread: a model trained on Shakespeare picks up patterns humans would miss — character-naming conventions, vague continental references a local wouldn't make, frequency of sexual puns, and how male vs. female characters start sentences. Any one or two of these tics is normal writing, but all of them together can't be faked. That's why such detectors are hard to fool.
Related event: Models can spot writing patterns invisible even to their authors(3 posts)→
More from Research
- Researcher Slams Journal Peer Review for Ignoring SOTA Bias-Correction Method ChromBPNet — anshulkundaje · 2026-09-23
- Stanford's Kundaje Calls Out TRACEBIND Paper for Ignoring ChromBPNet Baseline — anshulkundaje · 2026-09-23
- Academia operates like a content farm amid frontier-lab science disruption, researcher argues — RexDouglass · 2026-09-23
- EvolveTrade Paper: Evolving Agent System Prompts Beats Fixed-Prompt Trading Baselines — dair_ai · 2026-09-23
- MBZUAI x Ant Group open-source OPD work: IER selector helps thinking mode too — jiank_uiuc · 2026-09-23
- Sparse distillation paper: supervising just 0.1%-1% of tokens can match or beat full OPD — jiank_uiuc · 2026-09-23