GOFAI is basically "artificial intelligence" algorithms that don't involve neural networks or statistical learning. Deep Blue and Prolog (knowledge bases aka "expert systems") are your typical examples. They usually involve sophisticated tree construction and hand coded search algorithms. Modern ML also has tree stuff and symbolic algorithms but they are all heavily augmented by deep learning and probabilistic methods like Monte Carlo.
When people say Lisp was used for "AI research" and that it suffered from the "AI winter", GOFAI is what they are referring to. Neural networks exist back then but they were very small scale, good enough for some signal processing applications and as research curiosities but most of the powerful emergent capabilities were never truly realized until post millenium.
When people say Lisp was used for "AI research" and that it suffered from the "AI winter", GOFAI is what they are referring to. Neural networks exist back then but they were very small scale, good enough for some signal processing applications and as research curiosities but most of the powerful emergent capabilities were never truly realized until post millenium.