In 1965 the British mathematician Irving John Good published “Speculations Concerning the First Ultraintelligent Machine” in Advances in Computers, Volume 6 (Academic Press). Good was a statistician who had worked with Alan Turing on codebreaking at Bletchley Park during the Second World War, and the paper carries his affiliations at Trinity College, Oxford and the Atlas Computer Laboratory. It is the earliest clear statement of what is now called the intelligence explosion.
The paper’s central argument is compact. Good wrote: “Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an ‘intelligence explosion,’ and the intelligence of man would be left far behind.” Because the machine could recursively improve its own successors, he reasoned that “the first ultraintelligent machine is the last invention that man need ever make.”
Good attached a crucial condition to that optimism: the outcome holds only “provided that the machine is docile enough to tell us how to keep it under control.” That qualifier prefigures the modern control and alignment problem decades before it had a name. The first draft of the monograph was completed in April 1963 and the published version slightly amended in May 1964.
The intelligence-explosion idea later became a load-bearing assumption for much of the existential-risk literature, including Nick Bostrom’s 2014 book Superintelligence. The scanned reprint hosted at Virginia Tech’s statistics department resolves to the genuine Advances in Computers article; the ScienceDirect entry (DOI 10.1016/S0065-2458(08)60418-0) is the canonical publisher record, though its full text sits behind a paywall.