Defining intelligence in the Watson computer on Jeopardy!
The IBM computer Watson, appearing on Jeopardy! this week, may defeat the human players. But humanity is the real winner by gaining a better understanding of what real intelligence is.
Question: What is intelligence?
OK, so that probably won’t be a clue on the three special “Jeopardy!” shows this week, ones in which, yes, a computer named Watson – the size of 10 refrigerators, named after IBM’s founder – will compete against two humans, champion players Ken Jennings and Brad Rutter.
Still, the question “What is intelligence?” will be on the tongues of many “Jeopardy!” viewers as they watch this megacomputer use the latest in “artificial intelligence” (AI) and try to master the breadth of human knowledge and the depth of the human language.
This historic TV event could simply be seen as yet another milestone in the long quest to create machines as smart as, or smarter than, ourselves. We already live in an age of factory robots, voice-recognition software, complex Google searches, risk-lowering hedge-fund programs, military drones, and the Roomba vacuum cleaner.
In 1997, an earlier IBM computer, Deep Blue, beat chess champion Garry Kasparov. (“Well, at least it didn’t enjoy beating me,” the loser said later.) In the “Jeopardy!” contest, Watson will take a leap far beyond chess mastery – in which the rules are clear and only the math is complex. It will demonstrate an ability to learn from examples and mistakes. It will extract knowledge from terabytes of data, see patterns in language, assess the probability of being right, and even size up the human competition.
In test runs, Watson has beat out most contestants. But even if it wins this ultimate contest, the real winner may still be the humans. They will further understand that intelligence isn’t always, well, human, the kind that is presumably locked up in brain waves, emotions, and senses. After all, we have learned that other species have languages or use tools.
Since the 1950s, AI researchers have been forced to move beyond trying to replicate the human brain, realizing that intelligence is a bigger concept, one expressed in many aspects of life and that requires a process of growth. The best example? Watson itself is the latest expression of growth in AI, a machine that surpasses other computers in being nimble, subtle, empathetic, quick learning, even trustful – qualities found in any intelligent relationship.
Public fascination with humanlike computers has long been more in the fictional realm – Isaac Asimov novels, for example, or movies such as “The Matrix,” “Terminator,” or “Avatar.” More often than not these depictions evoke fear – think of HAL in the film “2001: A Space Odyssey.”
Fear has also been touted whenever complex machines have replaced human labor. But despite the Luddites’ complaint, a higher order of intelligence found in the advances of automation has only led to the creation of more jobs – intelligent ones, requiring humans to stretch their ability to manipulate symbols, synthesize ideas, or recall relevant facts.
Creating ever smarter machines, like Watson, is a way to expand human intelligence, not test its limits. They are tools not to replace us but to elevate our thinking.
What is intelligence? The answer lies in always looking for more answers, perhaps with better and better machines, like an endless game of “Jeopardy!