I. J. Good Defines the Ultraintelligent Machine and Articulates the Intelligence Explosion
In 1965, statistician I. J. Good, then at Trinity College Oxford, published 'Speculations Concerning the First Ultraintelligent Machine' in Advances in Computers, vol. 6, defining an ultraintelligent machine as one that surpasses all human intellectual activity and arguing that building such a machine would trigger an 'intelligence explosion.'
Background
By the early 1960s, most serious thinking about machine intelligence was either purely technical, focused on narrow problems like game-playing and theorem-proving, or it lived in science fiction. The two worlds rarely met. Researchers building programs that could play chess or solve algebra problems were not, for the most part, asking what would happen if those programs got very good. The question felt premature, or simply not scientific.
There were exceptions. Alan Turing had sketched the possibility of machines that learn and improve in his 1950 paper, and a handful of researchers had begun to think about whether a machine might one day match a human mind across the board. But matching was the ceiling of the ambition. The idea that a machine could exceed human intelligence, and then keep going, had not been given a precise form. It had no vocabulary, no logical structure anyone could pull apart and argue with.
Irving John Good was in a position to think about this carefully. A statistician who had worked with Turing at Bletchley Park during the Second World War and was by 1965 at Trinity College Oxford, he was comfortable moving between mathematics and large, open questions. He was also, crucially, willing to write down speculative arguments in a form rigorous enough to be checked.
What happened
In his paper published in Advances in Computers, volume 6, in 1965, Good set out a precise definition. An ultraintelligent machine, he wrote, is one that can far surpass all the intellectual activities of any man, however clever. That last phrase matters. He was not describing a machine that beats the average person at a specific task. He meant a machine that outperforms the best human minds across all intellectual work.
From that definition, he drew a logical chain. Designing better machines is itself an intellectual activity. So an ultraintelligent machine could design a machine better than itself. That new machine could do the same. Good called what followed an intelligence explosion. Each step in the chain produces a more capable designer, and there is nothing in the argument to make it stop. The whole sequence, he pointed out, would follow from building the first such machine, which meant humanity would only need to do it once.
Good was not predicting that this would happen soon, or that it would go well. He noted that the outcome would depend entirely on whether the machine had the right goals. The paper raised the danger; it did not dismiss it. But its main contribution was structural: it took an idea that had floated through fiction and gave it the skeleton of an argument. That made it possible to agree with, to refute, or to build on, in a way that no science-fiction treatment could support.
Why it mattered
Good's paper gave the concept of recursive machine self-improvement its first rigorous formulation, establishing the logical chain from a machine that can design better machines to an unbounded acceleration in machine capability. The term 'intelligence explosion' and the argument behind it became the foundational reference for later work on superintelligence and AI safety, influencing researchers including Vernor Vinge and Nick Bostrom. The paper also shifted discussion of advanced AI from science fiction into academic discourse, where it could be subjected to formal scrutiny.
People
Organisations
Trinity College, Advances in Computers
Sources
- Speculations Concerning the First Ultraintelligent Machine.Advances in Computers, vol. 6, Academic Press.Primary source
- Implications of a Software-Limited Singularity.Future of Humanity Institute, University of Oxford.Secondary
Cite this page
AI Achievements. (1965). I. J. Good Defines the Ultraintelligent Machine and Articulates the Intelligence Explosion. Retrieved 2026-08-22, from https://achievements.ai/milestone/first-ultra-intelligent-machine-defined-by-i-j-good
@misc{achievements_first_ultra_intelligent_machine_defined_by_i_j_good,
title = {I. J. Good Defines the Ultraintelligent Machine and Articulates the Intelligence Explosion},
author = {{AI Achievements}},
year = {1965},
url = {https://achievements.ai/milestone/first-ultra-intelligent-machine-defined-by-i-j-good}
}