1986

Backpropagation Popularized

David Rumelhart, Geoffrey Hinton, and Ronald Williams publish 'Learning representations by back-propagating errors' in Nature. The algorithm itself isn't new, but the paper finally shows it can train multi-layer neural networks to learn useful internal representations. The result is quiet — neural networks remain a backwater for another 25 years — but it is the mathematical engine that everything from AlexNet to GPT-4 will eventually run on. Without backpropagation, deep learning is impossible. With it, the road to modern AI becomes a matter of compute, data, and patience.