Where did AI start? — 1958: the neuron that learns
Today's AI is really just that one 1958 neuron, stacked up many, many times.
1 · Think first
In 1958, Frank Rosenblatt built a machine called the perceptron — an artificial neuron copying a single nerve cell in the human brain.
It does just one thing. It multiplies each incoming signal by a weight, adds them all up, and if the total clears a threshold it outputs 1 — if not, 0.
Here's what was really new — when it got something wrong, it nudged its own weights a little. It was the first machine that learned just from examples, without people writing the rules.
2 · Experiment · Weights and threshold
Try it — move the weights to clear the threshold
This artificial neuron decides whether to watch a YouTube video. Switch the signals on and off up top, and drag the weights below. If the total clears the threshold of 120 (the orange line), you get 1; if not, 0.
The Mark I Perceptron that Rosenblatt built in 1958 set these weights by turning motorized dials. Not a person — the machine itself, a little each time it got it wrong.
Figure
In one picture
Multiply, add, compare with the threshold — that's all one artificial neuron does.
3 · Quiz
Quick check
Three questions. Pick an answer and it's marked right away, with the reason why.
Today in one line
Today in one line
Today's AI is really just
that one 1958 neuron, stacked up many, many times.
Learn more
- The paper that first laid out a learning procedure that fixes the weights based on errors. Psychological Review 65(6), 386–408 — Frank Rosenblatt, The Perceptron (1958)
- 15 years earlier, the first paper to describe neurons as logic operations — the mathematical starting point of artificial neural networks — McCulloch, Pitts, A logical calculus… (1943)