After the euphoria of the 1950s, researchers thought they were on the verge of creating a complete artificial mind. But, as in any epic story, after the dawn came a period of technological darkness.

1. The Perceptron and the Death of Hope (1960s)

In 1958, Frank Rosenblatt introduced the Perceptron, the ancestor of today’s neural networks. The press at the time wrote that the military would soon have robots that “would walk, talk, and be aware of their own existence.” But in 1969, Marvin Minsky and Seymour Papert published a book that mathematically demonstrated the limitations of this model. It was the coup de grace that cut off almost all funding for neural networks for over a decade.

2. The Lighthill Report: “Nothing of value has been produced”

In 1973, the British government requested a report on the state of AI research. The result, known as the Lighthill Report, was devastating: it claimed that AI had failed to achieve its grandiose goals and that everything it had achieved could have been done by simple statistical methods. Funding dried up instantly in Europe and the US.

3. Expert Systems: A Breath of Air in the 1980s

AI survived through Expert Systems. Instead of trying to imitate an entire brain, researchers focused on ultra-specific areas: for example, a program that could diagnose blood diseases better than a resident doctor. This was the commercial “First Wave”, where companies began to see the value of the money invested.

4. The Second AI Winter

But Expert Systems also had a problem: they were difficult to maintain and could not “learn” on their own; they had to be programmed manually with thousands of rules. When the personal computer market exploded, these expensive systems became obsolete, and AI again entered a period of silence until the late 1990s.