NBER study: Pangram detector achieves near-zero error rates
soumitrashukla9 · x · 2026-08-19
An NBER working paper evaluates AI text detectors Pangram, OriginalityAI, GPTZero, and RoBERTa. Commercial tools outperformed open-source ones, with Pangram achieving near-zero false positive and negative rates on medium-to-long texts, remaining robust against short passages and 'humanizer' tools.
More from Research
- Classic Paper: The Surprising Creativity of Digital Evolution — Ghost_Pilot_MD · 2026-08-19
- HarmProfile Benchmark: Harmfulness and Diversity Rise with Model Capability — Zhouyuan Ma · 2026-08-19
- Why Removing the Vision Encoder Can Be Better — From an Infra Perspective — liuziwei7 · 2026-08-19
- Matmul Optimization Bottleneck: Data Movement, Not Multiplications — yaroslavvb · 2026-08-19
- Hot take: "Skill optimization" is just repackaged prompt optimization — IanArawjo · 2026-08-19
- Real AI risks lie outside the model: permissions, data, and presentation — bigdata · 2026-08-19