When we humans learn, we are gathering information that we add to our memory bank, changing incrementally our knowledge, skills, or attitudes. Programs that enable a computer to gather data to further advance its capabilities are said to produce “machine learning.” In this sense, machine learning is a reality.
For example, in the May 1997 contest between chess master Garry Kasparov and IBM’s Deep Blue, the two adversaries came to game six with the match all tied. Deep Blue, playing white, opened and Kasparov responded with what is called the Caro-Kann defense. By the eighth move, Deep Blue consulted its massive reference library on the Caro-Kann defense and offered a knight sacrifice that Kasparov accepted. Almost immediately after this knight capture, Kasparov realized his mistake and after 19 moves Deep Blue had emerged victorious, 3½ – 2½, in the battle between carbon-based and silicon-based brains. Deep Blue stored this entire game in its memory bank so that on future occasions if it met a similar situation, it would know how to proceed. This is a simple example of machine learning, but more sophisticated examples are emerging.
After the victory of artificial intelligence in defeating a chess master, many skeptics asserted that machines could never win a game like Jeopardy! that requires a sophisticated sense of metaphor–connecting ideas from different realms that are structurally related. By inputting information from the natural and social sciences, as well as literature, the arts, and sports into their computer named Watson, IBM attempted to “teach” Watson the associations linking disparate ideas connected in metaphoric ways. In 2011, Watson, soundly defeated Jeopardy! champions Ken Jennings and Brad Rutter.
Artificial intelligence continues to expand its scope by developing computers that are more “intelligent” than themselves. While this is still in its infancy, there are those who believe there is no limit to how intelligent computers will become. Others echo the skepticism of Noam Chomsky, who asserted, “Watson understands nothing. It’s a bigger steamroller. Actually, I work in AI, and a lot of what is done impresses me, but not these devices to sell computers.”
The bottom line is that machine learning is real, but enhancing machine learning can, without question, be a profitable enterprise.
