From Authorial Validation to Autonomous Truth Production: Artificial Intelligence and the Future of the Academic Paper

Artificial intelligence is breaking the institutional bargain under which an academic paper served both as a contribution to knowledge and as evidence of its producer's competence. When AI can supply literature search, analysis, objections, and prose—and may eventually perform most of a research cycle—paper quality becomes weaker evidence of what the named author or institution can do. That signaling failure does not discredit the paper's claims: their credibility still depends on evidence, methods, independent testing, and correction. Drawing together work on scientific incentives, research evaluation, reproducibility, AI governance, and autonomous science, this paper develops a two-function model and follows its institutional trajectories. Its central result is that verification capacity, not generation capacity, becomes the binding constraint. If AI expands candidate claims faster than independent checking, existing publication incentives produce credential inflation, reviewer overload, prestige-based triage, and accumulated error. If institutions separate personnel assessment from claim assessment and use AI to expand checking, the same technology can support replication, correction, living evidence records, and knowledge production beyond human authorship. The recommended settlement is to evaluate people through limited, auditable records of their actual contributions; evaluate claims through provenance and independent performance; reward verification; and raise safeguards with research autonomy. Experts and papers remain indispensable for judgment, explanation, responsibility, and governance, but neither expert status nor polished authorship should determine what is believed.