Marvin Minsky (August 9, 1927 to January 24, 2016) was an American mathematician and computer scientist who co-founded the MIT Artificial Intelligence Laboratory in 1959 and received the Turing Award in 1969. He is often called a "father of artificial intelligence", but that phrase is an informal label and no body awards it; the record supports specific contributions instead. The best known is Perceptrons, a book written with Seymour Papert that analyzed what simple one-layer neural networks can and cannot compute.
Key takeaways
- Minsky earned a Harvard degree in 1950 and a Princeton PhD in mathematics in 1954, and MIT credits him with building an early neural network simulator in 1951.
- He joined MIT in 1958, co-founded its AI Laboratory in 1959, and won the 1969 Turing Award.
- Perceptrons (1969) showed formal limits of single-layer perceptrons, with parity and the XOR function as the textbook case.
- Whether the book stopped neural network research is disputed; a sociological study calls that story an "official history".
- His broader legacy includes frames, The Society of Mind, the LOGO turtle, and an early confocal microscope.
Career in brief
MIT's obituary records that Minsky was born in New York City, graduated from Harvard with honors in 1950, and completed a mathematics PhD at Princeton in 1954. In 1951 he built what MIT calls the first neural network simulator. He joined MIT's Department of Electrical Engineering and Computer Science in 1958, and the Artificial Intelligence Laboratory, today part of CSAIL, was founded there in 1959.
The IEEE Computer Society's pioneer biography adds that he worked at Lincoln Laboratory from 1957 to 1958, directed the AI group of Project MAC from 1968 to 1970, and directed the AI Laboratory from 1970 to 1972. It credits his 1960s work on pattern recognition and vision with leading to "frames of information", a way of organizing stored knowledge that influenced knowledge engineering. He died in Boston at age 88.
Perceptrons: what the book actually showed
In 1958, Frank Rosenblatt had proposed the perceptron, a unit that adds weighted inputs and fires when the total passes a threshold. Minsky and Papert studied such units mathematically. The IEEE Computer Society biography says they analyzed Rosenblatt's perceptron and defined its mathematical limits. Their book appeared in 1969, according to its own introduction, though the bibliography of that biography dates it 1968, and an expanded edition followed in 1988.
The classic example is the exclusive-or function (XOR), which outputs one when exactly one of two inputs is on. The next section works through why a single threshold unit cannot compute it. The book's results are broader, covering which patterns a perceptron of limited connectivity can detect, but the XOR case shows the idea in four lines.
A worked example: why one unit cannot compute XOR
A single unit with weights w1 and w2 and threshold t fires when w1 times x1 plus w2 times x2 exceeds t. For XOR to work, four conditions must hold at once.
| Inputs (x1, x2) | Required output | Condition on the unit |
|---|---|---|
| (0, 0) | 0 | 0 does not exceed t, so t is at least 0 |
| (1, 0) | 1 | w1 exceeds t |
| (0, 1) | 1 | w2 exceeds t |
| (1, 1) | 0 | w1 plus w2 does not exceed t |
The second and third rows give w1 plus w2 greater than two times t. Because t is at least 0, two times t is at least t, so w1 plus w2 must exceed t, which contradicts the fourth row. No choice of weights works. Adding a second layer removes the problem: one hidden unit computing OR and another computing AND can feed an output unit that fires for "OR but not AND", and that is XOR. The mathematics is simple; the historical argument is about what the book implied for deeper networks.
Did the book end neural network research?
A common retelling says that Perceptrons caused funding for neural networks to collapse until the 1980s. Mikel Olazaran, in a 1996 article in Social Studies of Science, analyzed the perceptron controversy and argued that this official interpretation grew out of the rise and legitimation of symbolic AI, and that the late-1980s neural network revival led to the history being rewritten. The article does not say the book had no effect. It says the clean story of a single decisive blow is a later construction.
For readers, the safe summary is narrow. The book proved limits for specific, restricted perceptrons. Whether those limits discouraged researchers, and how much, is a question for historians, and the sources reviewed here do not settle it.
Other work that carries his name
MIT lists several contributions beyond neural networks. He co-invented the LOGO "turtle" with Papert, and the Muse synthesizer with Ed Fredkin. He invented the earliest version of the confocal scanning microscope. MIT calls his book The Society of Mind a seminal exploration of intellectual structure and function, and his last book, The Emotion Machine, appeared in 2006. His honors included the Japan Prize, the Benjamin Franklin Medal, and the BBVA Foundation Frontiers of Knowledge lifetime achievement award.
His name is also on the 1955 Dartmouth proposal that used the phrase "artificial intelligence", as one of four authors, a document covered separately on this site.
FAQ
Is Marvin Minsky the father of artificial intelligence?
The title is informal and is also given to others, including John McCarthy, his co-author on the Dartmouth proposal. The documented facts are the Turing Award in 1969 and his role in founding the MIT AI Laboratory.
What is a perceptron?
It is a simple model of a neuron: it multiplies inputs by weights, adds them up, and outputs one if the sum passes a threshold. Rosenblatt published the idea in 1958.
Can neural networks solve XOR today?
Yes. A network with one hidden layer can compute XOR, and modern deep learning systems use many layers. The limit applied to single-layer units.
What did Minsky win?
MIT's obituary lists the Turing Award (1969), the Japan Prize, the Benjamin Franklin Medal, and a 2014 Dan David Foundation Prize, among others.
Related reading
- Who is called the father of AI, and why
- How AI and cognitive science overlap
- Stuart Russell and modern AI
- AI in scientific research
More in AI Basics.
Last verified: October 2026
Sources
- Marvin Minsky, 1927-2016 (obituary), MIT News, January 25, 2016. source page
- Marvin Lee Minsky, Computer Pioneers, IEEE Computer Society.
- A Sociological Study of the Official History of the Perceptrons Controversy, Olazaran M, Social Studies of Science 26(3), 1996
- Perceptrons: An Introduction to Computational Geometry, Minsky M, Papert S, MIT Press, 1969; expanded edition 1988
- The perceptron: a probabilistic model for information storage and organization in the brain, Rosenblatt F, Psychological Review 65(6), 1958