GPTZero Glossary

An AI-text detector launched in January 2023 and widely used in education; its scores are signals, not proof of authorship.

GPTZero is an AI-text detector launched publicly in January 2023 by Edward Tian, then a Princeton undergraduate. The first version, built over the New Year holiday, drew tens of thousands of users within days. GPTZero later became a company serving education and other institutional markets.

For this Dictionary, GPTZero is one of the named players in the AI-detection economy critiqued in The Olang’ Trap. A detector estimates whether text resembles patterns associated with model output; it does not observe who wrote the text. Its score should therefore prompt further inquiry, never serve as proof of misconduct.

Early descriptions of GPTZero emphasised perplexity and burstiness: measures of textual predictability and variation. Detector methods have since changed, and results depend on the detector, model, text length, genre, and editing. A 2023 study of seven detectors—including GPTZero—found that they frequently misclassified essays by non-native English writers. That is evidence about a detector class under tested conditions, not proof that every current product behaves identically. It is enough to rule out automated accusation.

Sources

• South China Morning Post, “Who is Edward Tian?”, January 2023.
• Weixin Liang et al., “GPT detectors are biased against non-native English writers”, Patterns, 2023.

See also

• The Olang’ Trap — the structural critique
• Pangram Labs, Originality.ai — competitors
• The Sinceerly Stack
• AI Writing

Return to Dictionary All Entries (A–Z) For Students Other Writing Capstone 2.0