← Home
Cautionary Tales · June 15, 2023 · 40m
The Summer They Promised Us AI
Cautionary Tales
The 1956 Dartmouth Summer Conference — when the founders of artificial intelligence predicted that machines would match human intelligence within a generation. Harford examines why AI's most confident predictions have consistently been wrong.
Highlights
•
Technology predictions consistently overestimate short-term progress and underestimate long-term transformation
In 1956, AI pioneers predicted human-level AI within 20 years. In 1986, they predicted it within 20 years. In 2006, they predicted it within 20 years. The timeline keeps shifting but the confidence never wavers.•
Naming a field can constrain its development — 'artificial intelligence' created unrealistic expectations
Harford argues that calling the field 'artificial intelligence' committed researchers to mimicking human cognition rather than pursuing machine capabilities. The name shaped the research agenda for 70 years.