Marvin Minsky
Co-founder of the MIT AI Lab and a founding figure of the field — his 1969 book with Seymour Papert exposed the perceptron's limits and helped trigger the first AI winter.
Marvin Minsky (1927–2016) co-founded the MIT Artificial IntelligenceAI (Artificial Intelligence)AI is the field of building systems that perform tasks normally requiring human intelligence — reasoning, perception, language, and decision-making. Laboratory in 1959 and spent the next five decades arguing, in books, papers, and famously combative talks, about what intelligence actually is and whether it could be built from parts. This site is named for him.
His most consequential technical work is also his most controversial. In 1969, with Seymour Papert, Minsky published Perceptrons, a rigorous mathematical analysis of Frank RosenblattFrank RosenblattPsychologist who built the perceptron in 1958, the first neural network that learned its own weights from data rather than having them set by hand.'s single-layer perceptronPerceptronThe Perceptron (1958) was the first learning system built from an artificial neuron, and the direct ancestor of the neural network training loop.. The book proved these networks couldn't represent even simple functions like XOR — correct, and already understood to be fixable by adding hidden layers, but the book was read (fairly or not) as a broader case against neural networksNeural NetworkA neural network is layers of simple weighted-sum-plus-nonlinearity units (neurons) chained together, trained by gradient descent and backpropagation. entirely. Funding for connectionist research dried up for most of a decade, a period historians call the first AI winter. Minsky spent his later career mostly on symbolic approaches, developing "frames" — structured representations of everyday knowledge — and in 1986 published The Society of Mind, arguing that intelligence emerges from many simple, specialized processes acting together rather than one unified mechanism, an idea with clear echoes in how modern agentAgentAn agent puts an LLM in a loop with tool access, letting it decide autonomously which tools to call and in what order to accomplish a multi-step goal. systems are built.
He remained skeptical that scaling up existing methods, connectionist or otherwise, would be sufficient for general intelligence — a debate that is, in different language, still the live disagreement between scaling optimists and skeptics today.
See also: Frank RosenblattFrank RosenblattPsychologist who built the perceptron in 1958, the first neural network that learned its own weights from data rather than having them set by hand., John McCarthyJohn McCarthyCoined the term "artificial intelligence" in 1955, organized the field's founding 1956 Dartmouth workshop, and invented Lisp.
Learn more: History & Landscape · Wikipedia: Marvin Minsky
Mentioned in
Lessons where this comes up in context.
John McCarthy
Coined the term "artificial intelligence" in 1955, organized the field's founding 1956 Dartmouth workshop, and invented Lisp.
Frank Rosenblatt
Psychologist who built the perceptron in 1958, the first neural network that learned its own weights from data rather than having them set by hand.