Frank Rosenblatt

Frank Rosenblatt (1928–1971) was an American psychologist who built one of the first machines that could learn. Trained in the mind rather than in engineering, he set out to make a device that worked a little like a brain — and in doing so lit the fuse on the whole field of neural networks.

The defining work

In 1958 Rosenblatt built the Perceptron, a machine modelled on a single neuron: it took in signals, weighted them, and learned to adjust those weights until it could tell one pattern from another. It's the direct ancestor of the deep neural networks behind today's AI. The Perceptron could even learn to recognise simple shapes — an astonishing feat for the 1950s.

The press went wild, reporting that Rosenblatt's machine was the "embryo" of a computer that would one day walk, talk, and be conscious. Then rival researchers published a book pointing out sharp limits to what a single perceptron could do, and funding for neural networks dried up for years — the first "AI winter." Rosenblatt, also a keen astronomer, died young in a boating accident on his birthday. Decades later his core idea came roaring back, and now powers much of modern machine learning.