Artificial Intelligence (6.034)

It covers the classical breadth — search, constraints, rule-based reasoning, core machine learning — including the symbolic approaches that pre-dated neural networks. That history is exactly the intellectual context most modern, deep-learning-first courses skip, and it makes you understand why the field went the way it did.

It does not cover current deep learning or large language models, so treat it as foundations rather than the latest methods, and there is no certificate. Pair it with an applied ML or NLP course for the modern side. The recordings are older, but the fundamentals remain relevant.

MIT 6.034 with Patrick Winston is one of the most beloved AI courses in existence — Winston was the director of MIT's AI Lab for many years and taught this course until his death in 2019, leaving behind lectures that are simultaneously rigorous and genuinely funny. The course covers AI's classical foundations: search algorithms (A*, minimax), constraint satisfaction, rule-based systems, machine learning fundamentals, genetic algorithms, neural networks, and language.