Humans want proof of human, and interviews are tightening

A post on the Competitive Programmer's Handbook drew a comment asking if leetcode is still in interviews and saying 'we need to start gatekeeping this profession hard', a reversal after two decades of the opposite stance. Elsewhere, a ballot-security article lost a reader at the phrase 'load-bearing', with the comment 'Claude trauma is real'. A Samon puzzle post linked a Pangram AI-detection report, an Ask HN asked for a flag for AI-generated articles, and Cliff Stoll's 'rumors of my death' post involved AI-generated obituary content.

The pattern: when generation is free, people retreat to filters. Interviewers want hard algorithm tests again, readers want AI-writing tells, and communities want a flag. The unresolved bit is that detectors are unreliable and the tells are stylistic, so everyone is guessing.

The counterpoint, from a related HN thread: pushing too hard on AI stigma makes legitimate users unwelcome, and some voters don't notice or don't care.