Back Rohan Paul

Rohan Paul· @rohanpaul_ai · X· 2026-10-23 22:34 · Original publication time AI score20

Post criticizes treating AI text detector output as a final verdictMachine translation

Automatically verified and published · Generated and evidence-checked automatically; not reviewed by a human.

AI introduction

The post points out that some people demand checks on model output while treating detector output as final. A post it quotes also questions Pangram, arguing that it struggles to distinguish human writing edited with AI from text generated entirely from a prompt. That is the poster’s assessment, not a verified test result.

Original post · Original

The article text is unavailable in this language; an existing version is shown.

Rohan Paul@rohanpaul_ai

The same crowd that says never trust a model’s output without checking will paste a detector’s output as the final word. https://t.co/qK4cfoVBpZ

QuotePeyman Milanfar@docmilanfar

People treat Pangram as an infallible oracle. They weaponize its probabilities to shame others by some arbitrary purity standard. The irony is that it’s mostly redundant: obviously AI-generated stuff is already recognizable without it. And for more nuanced writing, Pangram fails to distinguish between text written by a human and edited by AI, vs text that is generated whole-cloth from a prompt. At the end of the day, @pangram ’s goal is about monetizing people’s anxiety about AI - it’s not about protecting creative integrity. They sell a false binary certainty to those gullible enough to ignore probabilities.

View this post on X ↗

22:34 · 2026-10-23 · UTC+8

Found an error? Send a correction