AI Is Causing Signal Collapse Everywhere All at Once

AI is rapidly weakening the shortcuts people use to judge quality, from polished writing and passing tests to online reputation and engagement with prior research. These signals worked partly because faking them used to require much of the effort involved in producing good work: a strong README accompanied real software maintenance, a successful assignment reflected learning, and a credible literature review required reading relevant studies. Generative AI sharply lowers the cost of producing convincing appearances, so genuine work, careless AI-assisted work, and deliberate deception can look alike. This change is unusually broad and fast, affecting open source, education, and academic peer review at once; earlier signal erosion, such as grade inflation or spoofed caller IDs, was generally slower or narrower.

Signaling theory predicts that when reliable signals disappear, people must develop new ones or pay more to evaluate work directly. If neither is practical, a “market for lemons” emerges: buyers cannot justify paying a premium for good work, strong contributors lose incentives, and low-quality offerings gain ground. That could mean avoiding open-source dependencies, struggling to assess graduates’ learning, or declining to review papers whose substance is hard to verify. The consequences include rising transaction costs and stalled collaboration; one possible fallback is reliance on smaller networks built on repeated interaction, personal trust, and reputational penalties. The central challenge is rebuilding efficient ways to distinguish quality without assuming that effortful-looking outputs prove genuine competence.