Sometime in the 40’s Turing and his friends began to talk of and even predict AI. It has been about 20 years off ever since. Aristotle said “All men are mortal; Socrates is a man; therefore, Socrates is mortal.”. Thus did Aristotle draw attention to something blindingly obvious yet in need of saying. For two thousand years some thought that human intelligence was defined by logic. (There were dissenters but I don’t think they contributed much.) I think that the discussions stimulated by Turing remarks centered mainly on logic buttressed by what mathematicians, such as Turing, were discovering about computers and logic.
I conjecture that humans, like advanced animals, recognize patterns for clear survival purposes. Even what lower animals accomplish here is impressive but I do not explore that here. Humans have speech and have come to contemplate what we say in addition to what we mean. This contemplation led to grammarians: the second oldest profession. Grammar is a collection of patterns found in speech. Just beyond grammar is a pattern of logic in some of our speech. The pattern of logic bears on what we believe and this was also noticed by Aristotle. Many noticed this before Aristotle, I am sure, but Aristotle noticed that it was important, said it, and wrote it, and it thence staid said.
People (and many animals) recognize patterns. Group theory can explain patterns and many thought that group theory is math which is logic and there was nothing that threatened the idea that intelligence was mere logic. Group theory is superb at describing patterns, but not much good at discovering patterns, such as logic.
Logic was the lamppost in the first couple of decades of AI. There was not the key they sought there but the programs they wrote to play chess then awaited only fast computers to defeat our chess champions. A few demonstrations that computers could process ideas were proven then too. The proof of one of Euclid’s early theorems was shortened.
Lowly animals recognize patterns defined in their DNA. Sometime since the Cambrian explosion animals learned to recognize new patterns first seen in the lifetime of an organism.
Neural nets somehow defined by DNA can recognize several patterns or at least react is useful ways to patterns. In desperation AI workers sought to use random collections of Boolean circuits to solve recognition problems. They invented back-propagation as an award like mechanism to encourage those circuits that helped succeed in recognizing some pattern. Their successes were good enough to support companies who produced such “neural nets”.
GOFAI was conceived to include the task of gathering propositions that describe our world. Two collaborators might meet and exchange propositions and thus extend the art. Neural nets had no such collaboration scheme; each net seemed to be a stand-alone artifice. Monica Anderson mentioned this first to me and again something that needed saying was said.
It is not clear the people recognize patterns better than other animals, but people convey discovered patterns in language such as “There are usually lions at the water hole in the morning.”. There are other uses of language but this use must have been an early adaptive force on the evolution of language.