Same text judged 100% AI and 100% human as AI detectors contradict each other
generativist · x · 2026-10-04
Writer DrPippaM tested her own human-written text across AI detectors: Pangram flagged it 100% AI while ZeroGPT called it 100% human. Pangram even offered to fix the result, which she declined, leaving the same text with contradictory extreme verdicts — and she's calling out the platforms. generativist mocked the absurdity with a VCR-porn joke, underscoring how unreliable detectors still get treated as evidence.
More from Fun
- Quadriplegic seller says Amazon's new Agent Policy blocks him from using Grok Bot to run his store — Baconbrix · 2026-10-04
- Agent Civilization: Why Rollout Agents Prefer a Half-Megatoken Reward Hack — voooooogel · 2026-10-04
- AI Circles Clash Over Consciousness: ctjlewis Slams Yudkowsky's 'Qualia' Talk — ctjlewis · 2026-10-04
- Users Praise AI Game Platform Spawn as First AI App That's Actually Fun to Use — TAbrodi · 2026-10-04
- I've swapped 47 open Chrome tabs for 47 open AI conversations — evielync · 2026-10-04
- GergelyOrosz: an LLM is as conscious as my screwdriver, sparking debate — dankaplan · 2026-10-04