AI Is Coming for Tasks Before It Comes for Jobs

AI progress is shifting from a "crashing waves" model of sudden, disruptive automation—characteristic of software engineering—to a "rising tides" phenomenon where capabilities gradually improve across diverse economic tasks. While inference costs are plummeting, fixed costs associated with data gathering, validation, and infrastructure remain primary barriers to adoption. Current models perform significantly better at "minimally sufficient" tasks than at "superior quality" work, leaving a critical role for human oversight in prompting, iteration, and risk management. As AI tools proliferate, the bottleneck for economic impact is moving from generation to verification and deployment. Consequently, maintaining human expertise through apprenticeship and training remains vital, as AI systems continue to reshape labor markets and demand new strategies for balancing automated efficiency with the necessity of human judgment in high-stakes fields like mathematics and cybersecurity.