How to Think About AI

Artificial intelligence (AI) technology isn’t new, but it’s now developing exponentially.

People imagined the advent of AI — technology that can perform tasks normally considered requiring human intelligence — as early as 1818. Functioning systems date back to the 1950s, when the pioneering computer scientist Alan Turing published a paper on the subject, “Can machines think?” The first practical AI boom came through expert systems in the 1980s, but progress stalled in the 1990s due in part to the rise of the web, which diverted focus toward more widely accessible technologies. A major leap occurred when machine learning emerged. Systems began to learn from vast data sets, enabling breakthroughs like IBM’s Watson — a system that could answer queries posed in natural language rather than code — and DeepMind’s AlphaGo — a program that bested the top human Go player in 2016.

Catalyzed by the invention of the transformer architecture and the 2022 launch of ChatGPT, new AI systems soon began to reveal capabilities that exceeded human ability. In 2024, AI scientists Demis Hassabis and John Jumper won the Nobel Prize in Chemistry after developing...