IBM on Jeopardy!: Are we smart enough to figure out what Watson means?

I well remember when IBM’s Deep Blue handily bested chess genius Gary Kasparov back in 1997.

I had set up a small chess set next to my computer. My dial-up modem was slow, and the sites carrying the game live were erratic. But I still felt a thrill at watching a milestone in the history of artificial intelligence unfold before me - and a nagging concern that the era of human dominance was now potentially in doubt.

Watching IBM’s Watson destroy his human competition on Jeopardy! tonight brought us another big step closer. Despite a few strange missteps (last night Watson erroneously answered “The 1920s” even though Ken Jennings had been buzzed as a loser for answering “The Twenties” a moment earlier) and some strangely precise dollar-amount bids, Watson ended night two of the competition roughly $25,000 ahead of his nearest human competitor.

In the video below, Dr. David Ferrucci explains how Watson far outdoes Deep Blue and is now gaining on us.

So Watson is not a really-fast computer hooked up to the Internet where it finds the answers to Jeopardy!’s questions. It is actually a self-teaching machine that learns from its mistakes and the mistakes of the people it is playing against. (I assume between last night and tonight, Watson had digested the fact that The Twenties and 1920s mean the same thing.)

Computers have long had the capacity to “outthink” us with raw computation. But Watson does far more. Not only can it answer questions but it understands the puns and the jokes underlying so much of Jeopardy!’s charm.

Watson is also another step down the path to what futurist Ray Kurzweil calls “The Singularity.” He borrowed the phrase from physics to describe such a massive leap in artificial intelligence that we simply cannot see beyond its event horizon. As computers become exponentially every few months, they will ultimately reach a point where they acquire our emotional intelligence as well as our intellect. What that will mean for us and for them simply cannot be foreseen.

I was chatting about with the clerk running the scanner at Meijer tonight. He’s a bright young man I have talked with before. His reaction was, “Why aren’t we hearing more about this? This is huge.”

He went on to say that he recently read an article where a professor proposed that we might end up downloading our intelligence into computers so that we could send them to colonize other planets since our human bodies could never survive the trip.

Not just machine immortality but computerized colonization of the universe?

Afrwe returning home, I was pondering the possibilities when I was quickly brought back to earth by an article in the February 5 issue of New Scientist. (Sorry about the paywall.) Visionary Lester Brown was interviewed about his fears of the coming “food bubble.” In addition to the danger of financial speculation driving up food prices, Brown is concerned that a growing number of countries are draining aquifers beyond their capacity to recharge.

Saudi Arabia has almost completely depleted a fossil aquifer. “The World Bank says that 15% of Indian people (175 million people) are fed grain produced through overpumping,” says Brown. China is doing much the same.

Rising food prices worldwide are already fueling political instability. As climate change further threatens harvests worldwide, Brown’s warning reminds us that the real issue isn’t how smart computers are becoming but how stupid we are.