Wikipedia:Pangram
 | This is an essay. It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article or a Wikipedia policy, as it has not been reviewed by the community. |
Pangram is an automated AI detector that is favoured by some editors for detecting AI-generated text. It splits the text into segments and gives a judgement on each segment, ranging from "Human"-written to "AI-assisted" or "AI-generated" (the resulting percentages correspond to the proportion of text determined to be that status, they are not probabilities). Pangram typically has low false positives and comparatively high false negatives. Below are rules of thumb regarding how best to use it.
- Pangram (and any automated detector for that matter) should not be the sole basis for an accusation that something is AI-generated/-assisted; it must be combined with human judgement (and other evidence if possible). Consider making a judgement yourself first before using it.
- It is less reliable for smaller texts; input as much text as you can. 50 words is typically considered the bare minimum.
- It works best on prose (i.e. not code, tables, instructions, or mathematical notation).
- Only copy the plain text (i.e. remove all wikitext, headers, references etc.)
See also
References
- ↑ Robinson, Matt (2 December 2025). "Do AI Detectors Work Well Enough to Trust?". Chicago Booth Review. University of Chicago Booth School of Business. Archived from the original on 4 May 2026. Retrieved 28 May 2026.
All three commercial tools kept false positive rates below 1 percent, with Pangram's the lowest—essentially 0 across most decision thresholds. False negative rates were higher, coming in between roughly 0 percent and 2 percent for GPTZero and between 2 percent and 4 percent for Pangram. Originality.ai's false negatives were higher still: between 10 percent and 40 percent, depending on the model.