Herbert A. Simon predicts machines will be capable of any work a human can do, within twenty years
In 1965, Herbert A. Simon of Carnegie Mellon University published 'The Shape of Automation for Men and Management', in which he predicted that machines would, within twenty years, be capable of performing any cognitive task a human could perform, a claim that became one of the most cited and scrutinised forecasts in the history of artificial intelligence.
Background
By the early 1960s, artificial intelligence had already produced some striking early results. Herbert A. Simon and Allen Newell, both working at Carnegie Institute of Technology in Pittsburgh, had built two programs that genuinely impressed the research community. Logic Theorist, completed in 1956, could prove mathematical theorems by searching through possible logical steps. General Problem Solver, which Newell and Simon described in a 1961 paper, went further: it was designed as a general method for working through problems by comparing the current state of affairs with a desired goal, and closing the gap between them step by step.
These were not toy demonstrations. They were serious attempts to model how human reasoning actually worked, and Simon believed they pointed toward something much larger. The field at this point was young and, by most measures, optimistic. Progress had come quickly. The gap between what machines could do and what people could do still looked very wide, but to those closest to the work, it did not necessarily look permanent.
What happened
In 1965, Simon published The Shape of Automation for Men and Management, a book addressed largely at economists and managers rather than computer scientists. Its subject was the coming effect of automation on work and organisations. But the book contained a claim that would follow Simon for the rest of his career: that machines would, within twenty years, be capable of performing any cognitive task that a human being could perform.
The prediction was not made carelessly or in passing. Simon had spent years thinking about how people make decisions, how organisations process information, and how the programs he and Newell had built related to human thought. His argument was that there was no clear boundary between the kind of reasoning a computer could do and the kind a person could do, only a difference of current capability. If that was right, then capability would eventually catch up. Twenty years was his estimate of how long that would take.
It is worth being clear about what the book was and was not. Simon’s work on decision-making in organisations stretched back to his 1947 book Administrative Behavior, and The Shape of Automation drew on those ideas. But the AI forecast was a distinct claim, not simply a restatement of his earlier management theory. The two threads ran through the same book, and they have sometimes been blurred together in later accounts.
Simon was also not alone in making bold predictions at this time. Marvin Minsky, John McCarthy and others in the field made comparably confident forecasts about the pace of progress. What distinguished Simon’s version was the precision of the timeline and the authority behind it: by 1965 he was one of the most credentialled people in the field, and the claim landed accordingly.
The twenty-year window he named would close around 1985. By that point, the gap between the forecast and reality was plain enough that funding agencies and critics could point to it directly. The consequences for the field were serious.
Why it mattered
Simon's prediction, made by a figure of considerable scientific authority who had already developed foundational AI systems such as Logic Theorist and General Problem Solver alongside Allen Newell, set a benchmark against which AI progress would be measured for decades. Its failure to be borne out by the mid-1980s contributed to the first 'AI winter', as funding bodies and critics used the gap between the forecast and reality to question the field's credibility. The episode established an enduring caution about confident timelines in AI research, a methodological lesson that remains relevant.
People
Herbert A Simon, Allen Newell
Organisations
Carnegie Mellon University, Carnegie Institute of Technology
Sources
- The Shape of Automation for Men and Management (review citation via JSTOR).Harper & Row / JSTOR.Primary source
- GPS, a Program that Simulates Human Thought (Newell & Simon, precursor work).American Psychological Association.Secondary
Cite this page
AI Achievements. (1965). Herbert A. Simon predicts machines will be capable of any work a human can do, within twenty years. Retrieved 2026-08-22, from https://achievements.ai/milestone/progress-in-ai-by-herbert-simon
@misc{achievements_progress_in_ai_by_herbert_simon,
title = {Herbert A. Simon predicts machines will be capable of any work a human can do, within twenty years},
author = {{AI Achievements}},
year = {1965},
url = {https://achievements.ai/milestone/progress-in-ai-by-herbert-simon}
}