Thursday, February 10, 2011

Kurzweil on Computers on Jeopardy

This coming Monday, 2/14/2011, a computer will compete against two human champions on the TV show Jeopardy. If the computer is able to understand the quiz show questions (which often use puns, metaphors, and other word-play), some artificial intelligence experts will consider it a great leap forward. After all, understanding complex language is one of the hallmarks of human intelligence. And if the machine can understand the question, it will surely win, as it's got a monstrous advantage over any biological brain when it comes to searching for answers through massive amounts of data.

Here's artificial intelligence wizard and futurist Ray Kurzweil predicting that the computer will triumph.



I've been intrigued by Kurzweil since I read his book The Singularity is Near (and even understood some of it). And I was impressed by his accessibility when I emailed him about the Buddhist perspective of consciousness.

2 comments:

Doug said...

Kurzweil's comment to you about the "common toaster" made me think of the SyFy channel's TV series Battlestar Gallactica, a more recent re-imagining of the series from the late 1970's. If you haven't yet, check it out; most episodes can be streamed via Netflix.

The story revolves around the consequences of machines achieving sentience, rebelling against humanity, and eventually nearly extinguishing the human race. The survivors are pressed into many variations of moral dilemma revolving around whether a "machine" that is indistinguishable from a human in any way other than a DNA test can or should be considered a sentient being, worthy of moral consideration. All the more complicated when said machines have only recently enacted a nuclear holocaust upon humankind: the survivors tend to have strong emotions on the matter.

"Toaster" is a derogatory slang for "Cylon" - the name of the machine race. :)

stuartresnick said...

The story revolves around the consequences of machines achieving sentience

We may live long enough to watch the debate in Congress over whether machines should get the vote. That one will be philosophically sticky.

No matter how indistinguishable the robot/computers get vs humanity, the fact remains that the machines will be able to replicate themselves in a more efficient way. (Kurzweil points out that once we make a machine as intelligent as we are, that'll be the last machine we'll have to make, since that machine will be able to design and produce all future machines.)

So once we give robots the vote, they'll simply overwhelm us by replicating, and then exploiting the power of "one 'man' one vote." Wouldn't this make universal suffrage for all beings (whether bio or techno) impossible in practice, even if philosophically imperative?

Notice that we already face a version of this problem around the world (i.e., any group with a high birth-rate will eventually dominate at the ballot box). We've been able to ignore this contradiction because humans are so much slower and less accurate in our efforts to reproduce ourselves. When computers can do so without friction -- creating their replicas quickly and flawlessly -- it'll be a different matter.