People prefer AI writing in blind tests—until they're told it's AI

ipeirotis · x · 2026-09-10

Summarizing recent literature on AI writing, Panos Ipeirotis notes that under blind conditions people cannot reliably distinguish AI text from human writing, and often actually prefer the AI version. But once told a text was AI-generated, their preference flips to the "human" version—suggesting aversion to AI writing is driven by labeling bias rather than actual quality differences.

Related event: Blind Tests Show People Can't Distinguish AI Writing—Until Labels Bias Them(2 posts)→

Original post →

More from Research

Research channel →