# The degree's signaling function splits, and the split is regressive (2025 assessment)

**Claim:** The bachelor's degree performs two labor-market jobs on different clocks. Proof-of-learning (transcript, GPA) loses informational value by the early-to-mid 2030s, as AI-assisted coursework makes output unverifiable and AI-scored work-sample assessment undercuts it. Proof-of-sorting (admission selectivity) is a positional fact AI cannot counterfeit and persists to ~2045.

**Consequence:** Signaling decay does not shrink the university system evenly — it shrinks the nonselective middle and bottom, where the transcript was mostly certification, while raising the relative value of selective flagships, where the degree was mostly sorting. Predicted decomposition by 2035: a stable-or-widening selectivity premium and an attendance premium compressing toward zero.

**Leading indicator:** Whether large employers' skills-based-hiring commitments (IBM 2016; state policy 2022–24) convert into observed hiring mix by ~2030. Announcements are not the signal; conversion is. Expect a lag, since degree requirements persist as liability shields after they stop carrying information.

**Falsifier:** If skills-based hiring plateaus and degree requirements quietly return by ~2030, the signaling-collapse thesis fails and degree value holds through 2045 across the whole selectivity distribution.