The project of building machines that can perform tasks requiring human-like intelligence — from its founding in 1956 through symbolic AI, expert systems, and multiple winters, to the deep learning revolution of the 2010s and the large language model era — now the most rapidly developing technology in human history and the subject of the most consequential unresolved questions about the future of work, knowledge, and civilization.
Every decade since 1956 has been the decade AI finally arrives. Every decade has also been the decade AI disappoints. The difference this time — since roughly 2012, accelerating sharply since 2022 — is that the systems are actually doing things that were supposed to be impossible: passing bar exams, diagnosing cancer, writing code, generating images indistinguishable from photographs. Whether this is the transformation, or another prelude to disappointment, is the most consequential open question in technology.