Proof of human work is becoming a product problem

A UC Berkeley math professor says AI use leaves 'no point' in grading homework. Commenters shrugged and said the fix is old: pen-and-paper, in-class or oral exams, since homework only ever proxied for those. A Nikon photography contest winner lost the title after scientists spotted signs of AI, and the video about it was pulled from YouTube. The 'Ask HN: Add flag for AI-generated articles' thread argued that HN's new guideline against generated text covers comments but not submissions. Detection tools are unreliable, voters often can't tell, and a flag could stigmatize people who use LLMs lightly. The same worry shows up in 'I refuse to accept a censored Internet,' where a commenter asks how you even find real content through 'slop brigades.'

The through-line: every institution that relied on effort as a signal, from classrooms to photo contests to forum voting, is moving verification offline or into the room. Even the Nuremberg knight's tool book post drew a comment that AI images and writing made the author less trustworthy.

The nuance is that nobody has a reliable detector. The working answers are social and procedural, like in-person exams and human-only rules, not technical.