Extreme Cognitive Assistance and Open Futures

We owe young people special moral consideration in virtue of their developmental immaturity: we must not foreclose their futures by impeding the development of capacities they need for autonomous agency. I argue that providing children with extreme cognitive assistance—help that is ubiquitously available, very domain general, and largely substitutive of their own cognitive effort—violates this obligation. When children routinely rely on such assistance, they systematically fail to exercise core regulatory skills (inhibitory control, working memory, cognitive flexibility, and planning), and capacities that go unexercised predictably fail to develop normally. This matters urgently: building artificial systems capable of providing exactly this kind of assistance to children is the explicit, stated goal of the world’s largest technology companies, who are investing hundreds of billions of dollars toward achieving this aim. Worse, they may already have succeeded: cognitive assistance need not be maximally substitutive, available, and domain general to impede development. I conclude by considering whether compensating demand for cognitive exercise might arise elsewhere and argue that the prospects are dim.