user-pic

Watson and the Fear Factor

The NOVA episode Smartest Machine on Earth chronicles the four-year-long effort of a team of computer scientists at IBM to build a machine named Watson (after IBM's founder) that can play Jeopardy! -- a TV quiz show that represents for many the essence of human intelligence.

Working on this episode has caused me to reflect on a time many years ago when I made a NOVA called "Mind Machines." That film, like this new NOVA, is a reflection on the quest to create machines that can think like we do - in other words, artificial intelligence. In those days, that quest was just beginning and the results were pretty primitive. My film demonstrated a machine that could understand natural language well enough to manipulate different colored objects in a small world of blocks. It showed Eliza, a computer program that could respond like a psychiatrist, but was really just filling in the blanks based on what a "patient" had just typed in. The film began with a clip from Stanley Kubrick's classic "2001: A Space Odyssey," with its unforgettable scenes of the engaging but psychotic computer HAL running amok.

Artificial Intelligence, or AI, back then inspired two kinds of reactions: faith that it could be done and fear that it would. Most of the computer scientists I consulted were among the faithful. One accused me of being a "human chauvinist pig" when I expressed doubts. On the other hand, a few experts and most ordinary people worried that smart computers, like HAL, would get out of control and start running the world. What would happen to human values, these people asked, if silicon brains were more powerful than our own?

A lot has changed since then. First, a little thing called the Internet has come roaring into our lives. Unlike anyone I knew at the time, all the experts at MIT, Stanford and Carnegie Mellon that I interviewed had computers on their desks and could communicate with each other through cyberspace on a system called the ARPANET. A few years later, it's not just the experts who are benefiting from these innovations. Computers have become relatively cheap and accessible. Many of us have not one but several of them. We can communicate with each other across time and space and have a world of virtually limitless information at our fingertips.

And that's why, I believe, the fear factor has declined. We're used to computers. We know how dumb they are. We understand that they are limited to a narrow range of tasks we program them to do. The versatility of small children - the way they pick up language, navigate around obstacles, and learn to play games - is still out of reach for machines. And that's why, when we watch a computer named Watson compete against Jeopardy! champions, our admiration is less for its silicon brain than for the human ones that labored so long and hard to create it.

Publicist's note: Smartest Machine on Earth will premiere Wednesday, February 9 at 10pm on most PBS stations. Please check your local listings to confirm when it will air near you.

User Comments:

My fear isn't of intelligent computers, but the corporations that will wield them. The computers are just tools, but it's the corporations driven by profit within the capitalist system motive that will use them to eliminate workers without compensation, to outsmart consumers, to develop market trading systems that will become money machines which transfer wealth from human individuals to "investment corporations", etc. and all of this is already happening right now without systems like Watson.

It was very meaningful to see the IBM Thomas J. Watson Research Center last night on WNET. That is where I was fortunate to have worked as a patent attorney for 19 years. The wonderful 801 building designed by the great Eero Saarinen was displayed magnificently and the brilliance of the IBM scientists was there for all of us to see and hear.

Thank you so much.

Is this a completely fair test of a machine vs. a human? According to the show Watson is fed the answers via text message, whereas a human recives the answers visualy and auditorialy. It seems to me this gives the machine an unfair advantage. A human player must read and listen to an answer before begining, at least partially, to postulate an answer, then press a button to chime in before giving the question. Has the physicallity of the game been taken into account? Maybe a lag time before Watson sends an electronic signal? It would seem more balanced if Watson had to read and hear an answer and then press a button, possibly using a human analog (robot).
I've watched and enjoyed your show for many years,keep up the good work.

I agree Bret - I don't think it is fair, unless there is a 2-3 second lag time. Moreover, what happens with answers that are visual / auditory in nature? Since Watson cannot process those, those answers will never be part of the contest.

Fascinating program, just like 90% of all Nova programs. I still live without a DVR and if I miss a show, I can watch it online. If that was not the case, I would have bought a DVR only for Nova / Nature / Frontline.

Anyway, one of the things that irks me is that you cannot watch Nova shows online from anywhere outside of US. Am I right? I have friends in India, Australia, Europe and whenever I send them a Nova link, nothing shows up. Recently, I shared links on 'making stuff' series and everyone complained that they could not see anything.

I know there is something about intellectual property here, but PBS is losing an enormous opportunity of getting the word out about itself and may be even raise funds (if allowed by law) from international viewers. And moreover, PBS has always prided on sharing knowledge everywhere, why limit it to US?

If AI systems can learn to improve how they conduct their machine learning, maybe their only limitation will be their hardware. Watson has to depend on humans to tell it when it is right or wrong, which is the controlling force in its machine learning. Paula Apsell, you definitely created a great TV program here!

Bernard Schuster
Arrive2.net
Twitter.com/arrive2_net

I understood from the Nova Program that Watson had to deal with the voice recognition of the question. Yet Alex stated, at the beginning of the show, Watson would receive the statement via text message. Which is correct?
And if there were crashes and retaping during the show, what does this say about the real time aspect?
Ted

blog comments powered by Disqus