How AGI Could Reshape Society

AGI capable of generating and testing hypotheses across disciplines could compress decades of scientific work into years. Early indicators — AlphaFold's protein structure database, AI-driven materials discovery, autonomous chemistry labs — already show meaningful acceleration.

A science-capable AGI could shift the bottleneck from cognition to physical experimentation, regulatory approval, and ethical oversight.

Personalized AI tutors, available to every learner at near-zero marginal cost, could approximate the gains Bloom documented for one-on-one tutoring (the '2-sigma problem'). The shift from credentialing to actual capability development would be substantial.

The risk: cognitive offloading. Heavy reliance on AI for thinking tasks may degrade the underlying human skills, especially in developing minds.

AGI advisors could improve policy analysis, simulate intervention effects, and surface trade-offs invisible to human analysts. The same capabilities, deployed adversarially, could enable unprecedented manipulation of public discourse.

Democratic institutions were designed for human-scale information environments. AGI-era governance will require new mechanisms for legitimacy, accountability, and oversight.

If machines exceed humans at most cognitive tasks, the basis of human identity shifts. Historical analogues — industrial mechanization replacing physical labor — produced both flourishing and dislocation. AGI is likely to produce both at vastly greater scale.