Does Intelligence Have a Future?


In 1825, Johann Maelzel toured the United States exhibiting what appeared to be a mechanical device capable of playing chess. Although there were a number of essays written about this remarkable device, the most famous, “Maelzel’s Chess Player” was written by Edgar Allen Poe. Poe’s essay is of literary importance because it shows his development of an analytic method that later appeared in his stories of detection.

At the same time, Poe made some logical errors, including assuming that if it’s possible to build a machine that would win most chess games it would be easy to build a machine that would win every game. So far no computer has managed to win every game against a world champion, although IBM’s Deep Blue did win a six-game match against Garry Kasparov by a score of 3 ½ to 2 ½.

In 1899 Ambrose Bierce published a short story “Moxon’s Master” which dealt with a chess-playing machine. Bierce, who was also known for The Devil’s Dictionary, envisioned a chess-playing machine that could lose a game, but could not stand losing. “It gave me the impression of a disordered mechanism which had escaped the repressive and regulating action of some controlling part – and effects such as might be expected if a pawl should be jostled from the teeth of the ratchet wheel.” Maelzel’s automation drew crowds wherever it was shown while Bierce’s story generated debates about the limits of artificial intelligence long before AI was even a serious concept.

Now we’ve reached a level of AI that may approach Bierce’s fears. This past January, Carnegie Mellon’s computer program “Libratus” whomped a team of human poker champions. The champions and Libratus played 120,000 hands of Heads-up, No-Limit Texas Hold ’Em and the results was the computer beating the pros by $1,766,250 in chips (the actual prize was $200,000 in real money). The size of the win is important because it demonstrates that the victory was more than simply good luck.

By now we’ve become terribly jaded by the remarkable things that can be done with computers. Those of us who never mastered the intricacies of the Commodore 64 seem perfectly comfortable with the fact that the computer can beat a chess grandmaster, or two Jeopardy champions. Still, the fact is that these are (forgive me for even thinking this) relatively mundane victories that are achieved with nothing more than speed and power. Chess is an incredibly complex game with millions, billions of possible moves. One web site says that the number of possible moves is infinite, but since nobody can count up to infinity the estimate seems like overkill. Bigger and faster are clearly an oversimplification, and the ordinary desktop computer has no more relationship to IBM’s Watson than Curious George has to King Kong, but one does give you something of an idea of the other.

The unique feature of poker is that it’s a game played with incomplete information. Chess and checkers are honest, straightforward games with all the information laid out on the board. Even Jeopardy is essentially honest in the sense that, if a question is asked (or in Jeopardy the answer is given) there’s a basic rule that it has an answer. In contrast, poker is a game where bluffing is routine, where critical information is kept secret and the ability to interpret the facts can be a matter of considerable profit or tragic loss. The lessons learned in poker have their place in every-day life, from dealing with a used car salesman or real estate agent to international negotiations where a Middle Eastern strongman might offer a bluff that he has weapons of mass destruction. Politicians routinely make promises that can’t be kept, or distort the facts about policies that have been successful in order to denigrate the successes of their opposition.

It’s hard to say when the lessons of Libratus will be integrated into our daily lives. One of the more remarkable things about artificial intelligence is that these computers teach themselves, learning by trial and error, in the case of games playing computers by playing games against themselves and learning what works and what doesn’t. Right now, IBM is marketing Watson, the Jeopardy-playing computer, for any number of commercial applications. Their claim is “Watson is a cognitive technology that can think like a human. With Watson, you can provide personalized recommendations by understanding a user’s personality, tone, and emotion.” Prof. Stephen Hawking has famously warned that the creation of powerful artificial intelligence will be “either the best, or the worst thing, ever to happen to humanity.” In that speech Hawking also said “we spend a great deal of time studying history, which, let’s face it, is mostly the history of stupidity. So it’s a welcome change that people are studying instead the future of intelligence.”

If, that is, we still have a future.

Sam Uretsky is a writer and pharmacist living in New York. Email

From The Progressive Populist, March 15, 2017

Blog | Current Issue | Back Issues | Essays | Links

About the Progressive Populist | How to Subscribe | How to Contact Us

Copyright © 2017 The Progressive Populist

PO Box 819, Manchaca TX 78652