#Deep Learning History

The personal story of Hinton, LeCun, Bengio, and their collaborators during the years when neural network research was marginalised, underfunded, and dismissed. The decades of intellectual courage, the specific ideas that kept the research programme alive, and the slow accumulation of results that eventually made the field pay attention. How the most important research programme in modern AI survived by going underground.

The most important algorithm in modern AI was discovered at least three times before anyone paid attention. The full story of backpropagation — its multiple independent inventions, the decades it spent in the shadows of the symbolic AI winter, and the specific confluence of people, ideas, and timing that produced the 1986 paper that changed everything. How an algorithm refused to die, and what happened when the world finally listened.

He built the Perceptron — the first machine that could learn. The press called it a thinking machine. Marvin Minsky called it a dead end. History called Minsky right for fifteen years and then spent the next fifty years proving Rosenblatt correct. The most underrated figure in the history of AI, whose ideas now run the world.

He invented LSTM — the architecture that enabled speech recognition, machine translation, and early language models. He has been arguing, loudly and persistently, that he deserves far more credit for modern AI than he receives. The case for why he might be right — and the complicated personality that has made it harder to make that case effectively.