The General Problem Solver, or GPS, was a program created from 1957 onward by Allen Newell, J. C. Shaw, and Herbert Simon, the same trio behind the Logic Theorist. Where the Logic Theorist proved theorems in symbolic logic, GPS was an attempt at something more ambitious: a single program that could in principle attack any well-defined problem, from logic puzzles to algebra to arranging a sequence of moves, by the same general method. Their early account, “Report on a General Problem-Solving Program,” was presented at the 1959 International Conference on Information Processing in Paris.
The central idea GPS introduced was means-ends analysis. The program looked at the difference between its current state and the goal state, picked an operator that would reduce the most important difference, and applied it, recursively setting up sub-goals when an operator could not yet be used. In plain terms, it worked the way a person might plan a trip: notice the gap between here and the destination, choose an action that closes the biggest part of that gap, and repeat. This explicit reasoning about goals and differences was a genuinely new way to organize a search for a solution.
GPS also embodied a deeper claim that Newell and Simon made about the mind itself. They argued that human thinking and machine problem solving were two instances of the same thing, the manipulation of symbols, an idea they later crystallized as the physical symbol system hypothesis. GPS was meant not only as a useful program but as a working theory of how people solve problems.
Why business readers should care: means-ends analysis, separating a goal from the current situation and choosing actions to close the gap, remains a foundational pattern in automated planning and is echoed in how modern AI agents break a task into steps. GPS also illustrated the limit that haunted symbolic AI: a method general in theory still needed each problem hand-translated into its formal vocabulary, so the “general” solver was never as general as its name promised.