AI, the Peacock’s Tail, and Bullshit Jobs

Summary

The article argues that AI’s productivity gains may produce a paradox: when nearly everyone can reach a competent output level, the output loses its value as a signal of talent. Education, graphic design, polished writing, and other activities can serve partly to demonstrate intelligence, care, reliability, or organizational status rather than to create direct value. Before AI, differences in talent were more closely reflected in differences in output; generative AI weakens that connection by allowing less-skilled workers to produce work at roughly the system’s level. The author suggests that degrees could lose some signaling cachet if AI-assisted cheating becomes widespread, especially where institutions do not actively enforce standards. The article also describes a decline in the novelty and persuasive power of AI-generated images: once such images become commonplace, their use may signal limited effort or resources instead of quality. In the short term, AI may increase low-value signaling tasks, including compulsory training, compliance exercises, and self-promotional workplace activity. Drawing on David Graeber’s concept of “bullshit jobs,” the author argues that some of these tasks may eventually stop paying off and disappear as people and organizations find other ways to demonstrate value. These are presented as an argument about possible long-run effects, not as a measured forecast.