A photographer posted a sunset image after three hours of waiting for the right light. Within minutes: "Obvious Midjourney." "AI-generated garbage." "Nice prompt, bro." She posted behind-the-scenes footage proving she took the photo. Most accusers moved on without acknowledging the error. This is now the default reaction to anything that looks too polished online.
The Detection Problem
The tools people trust to identify AI content are unreliable at a fundamental level. Studies show specialized detection systems misclassify human-written text as AI-generated roughly 35% of the time, with false positive rates spiking on technical writing, non-native English, and creative work with distinctive styles.
The reason is structural, not fixable with better algorithms. Language models produce text by predicting which words humans would likely use in sequence. The better they get at that prediction, the more their output resembles human writing statistically. Any signature that flags "AI" will also flag certain patterns of human writing. The detection problem doesn't get easier as models improve. It gets harder.
Non-native English speakers and writers with unconventional styles bear the cost disproportionately. Their writing patterns overlap more with LLM output distributions -- not because they used AI, but because both produce text that deviates from the narrow band of "expected" native English.
The Social Incentive
Calling something "obviously AI" has become social currency. It signals technical sophistication and critical thinking regardless of accuracy. The accusation carries no penalty for being wrong, but delivers immediate status for being right -- or appearing right.
This creates an asymmetric incentive. Legitimate concerns about AI misuse and attribution get misdirected toward genuine human work. Creative communities increasingly require "proof of humanity" -- process documentation, behind-the-scenes footage, metadata -- that disadvantages creators with limited time or resources.
The Actual Trade-off
The question isn't whether AI detection matters. It's which error is more costly: occasionally failing to identify AI-generated work, or routinely dismissing human creativity as artificial.
False dismissal devalues real work, erodes trust in digital spaces, and creates barriers for new creators who lack established reputations. A hair-trigger accusation reflex produces more damage than the AI content it claims to police.
Before commenting "nice prompt" on someone's work, consider the possibility that you're looking at three hours of patience and a trained eye -- not a text box.