Reading I. J. Good's 1964 "Speculations Concerning the First Ultraintelligent Machine"
Reading Speculations Concerning the First Ultraintelligent Machine
AI summary
A LessWrong author closely reads I. J. Good's 1964 "Speculations Concerning the First Ultraintelligent Machine" and tests its predictions against today's AI progress. The author holds that he foresaw large artificial neural circuits that learn from experience through reinforcement, and that the premise "the machine is docile enough" corresponds to today's route of having AI research AI alignment, but he did not anticipate that high throughput would arrive before high intelligence. The AI Futures Project now expects an intelligence explosion to begin in 2027.