When output is cheap, verification becomes scarce: The outlook for AI in science

AI is increasingly taking on parts of the scientific method: reading, generating hypotheses, analysing, experimenting and deciding what should be tested next. The publication shows why, as output becomes cheap, verification, reproducibility, provenance and human judgement become scarce strategic resources. It connects AI loops and knowledge graphs with an integrated management framework for roles, decision rights, evidence standards and accountability—grounded in open science, with DeSci and blockchain used only where they measurably improve trust, incentives or governance.