Herbert Alexander Simon (1916–2001) was an American polymath whose work spans economics, cognitive psychology, political science, organization theory, and computer science. He is one of the founding figures of artificial intelligence, the originator of the concepts of bounded rationality and satisficing, and a co-creator of the first AI programs. He is one of the few researchers honored at the highest level in two distinct fields, receiving the ACM Turing Award in 1975 (with Allen Newell) and the Nobel Memorial Prize in Economic Sciences in 1978.
Background
Simon spent the bulk of his career at Carnegie Mellon University (and its predecessor, the Carnegie Institute of Technology) from 1949 until his death in 2001, where he helped build both the Graduate School of Industrial Administration and the School of Computer Science.
The ACM Turing Award (1975), shared with Allen Newell, was given for "basic contributions to artificial intelligence, the psychology of human cognition, and list processing," and for advancing the idea that computer science is an empirical inquiry into symbol manipulation. The Nobel Memorial Prize in Economic Sciences (1978) was given "for his pioneering research into the decision-making process within economic organizations."
Contributions to artificial intelligence
With Allen Newell and Cliff Shaw, Simon built the Logic Theorist (1955–56), widely described as the first running artificial-intelligence program; it proved theorems from Whitehead and Russell's Principia Mathematica. The trio followed it with the General Problem Solver, an attempt at a domain-general reasoning architecture. Simon and Newell presented this work at the 1956 Dartmouth Summer Research Project that gave the field its name.
The physical symbol system hypothesis, advanced with Allen Newell in 1976, holds that a physical symbol system "has the necessary and sufficient means for general intelligent action." This hypothesis framed the symbolic, or "good old-fashioned," program of early AI and is a standing reference point in debates over what general intelligence requires.
Simon is also frequently cited for the over-optimism of his early forecasts: in the late 1950s and 1960s he predicted, among other things, that machines would within twenty years be capable of any work a person could do. The gap between that prediction and what followed is a standard cautionary example in discussions of AGI Timelines and the recurring cycle of AI enthusiasm and AI winters.
Concepts
- Bounded rationality — the claim that real decision-makers, human or artificial, operate under hard limits of information, time, and computational capacity, and therefore cannot optimize in the way classical rational-choice models assume. It is a direct ancestor of how contemporary work reasons about the limits of both human and machine cognition.
- Satisficing — a term Simon coined, blending "satisfy" and "suffice": under bounded rationality, agents do not search for an optimal option but accept the first one that clears a "good enough" aspiration threshold.
Simon treated organizations as information-processing systems with limited rationality, a framing that is a precursor to current sociotechnical accounts of how AI systems behave inside institutions rather than as free-standing optimizers. He is cited in AI as Social Technology and on Henry Farrell's page, and is named in the Knight Columbia Democratic Matrix source as part of the intellectual lineage on institutions, organizations, and bounded decision-making.
Relationships
- depends-on: (intellectual lineage) institutional and organization theory; cognitive psychology
- related: AI as Social Technology, AGI Timelines, AI Winter, Henry Farrell, Cosma Rohilla Shalizi
Sources
No dedicated foundational source page yet. Page built from well-established biographical and historical record; named across AI as Social Technology, Henry Farrell, and Building AI for the Democratic Matrix: A Technical Research Agenda for Normative Competence and Normative Institutions (Hadfield + Trivedi + Hadfield-Menell, Knight Columbia, March 3 2026).
Page created 2026-05-25 by the gap-identifier (gap type 3 — a foundational AI-history figure referenced across multiple pages with no entity page). Confidence high — uncontested historical record.