A Personal Tribute to Patrick Henry Winston

Patrick Henry Winston was, by all standards, a rock star in the field of Artificial Intelligence. In 1970, Patrick wrote his Ph.D. thesis, in which he explored -- under the improvisational supervision of his advisor, Marvin Minsky -- the theoretical difficulties of learning, and wrote in Lisp a blocks-world program that could perceive blocks and block-enabled architectures (e.g. That computer program was able to learn to generalize its existing knowledge when comparing a baseline example architecture with a new example, and specialize its existing knowledge when comparing a baseline example with a near miss. That was the first effort ever in making machines learn things in ways that resemble how humans learn things. Some say that was "real" Machine Learning, much unlike statistical Machine Learning and neural-net Machine Learning, whereby programmers would program their computers to slavishly crunch through hundreds of billions of data points, which is nothing like how people learn new things, but has become popular because the theory behind them are much more understood and much easier to implement, and because this kind of big-data crunching is practically allowed for due to the tremendous computing power that we have today.