The concept of Intelligence Explosion is pivotal in discussions about the future of artificial intelligence and the potential path towards artificial general intelligence (AGI). It is based on the idea that an AI system, once reaching a certain threshold of intelligence, could continuously and autonomously enhance its own design or algorithms, leading to exponential growth in its capabilities. This process could theoretically result in the creation of an AI with intelligence far beyond human levels, posing both incredible opportunities and significant risks. The central concern is the alignment problem: ensuring such a superintelligent AI's goals are aligned with human values and interests.
The term "Intelligence Explosion" was first introduced by I.J. Good in 1965, in his paper "Speculations Concerning the First Ultraintelligent Machine". Good proposed that if a machine could surpass human intelligence, it would be capable of improving its own design in ways unforeseen by its creators, leading to an unprecedented acceleration in intelligence.
I.J. Good, a British mathematician and cryptologist, is credited with the foundational idea of the Intelligence Explosion. Since then, the concept has been further explored by numerous scholars and writers in the field of AI, including Eliezer Yudkowsky and Nick Bostrom, who have significantly contributed to discussions about the implications and ethical considerations surrounding superintelligent AI.