Sunday, January 13, 2008

RJS on TVO's The Agenda on Monday



I'll be a guest on TVOntario's flagship current affairs program The Agenda tomorrow, Monday, January 14, 2008. The Agenda is seen across Ontario at 8:00 p.m. and 11:00 p.m., and also at 5:00 a.m. Tuesday morning, and will be available online. The host of the agenda is the always insightful Steve Paikin.

We actually recorded this segment on December 11, 2007; the topic is roboethics (the ethics of robots), growing out of my editorial in the November 16 issue of Science, the world's leading scientific journal.

Above: science-fiction writer Robert J. Sawyer during an earlier appearance on The Agenda.



The Robert J. Sawyer Web Site


4 Comments:

At January 15, 2008 12:41 AM , Blogger RobertJSawyer said...

Just watched this. Man, I love doing interviews for TVO! Almost 20 minutes, in-depth coverage of the topic, great questions from Steve Paikin, nice beauty shots of my book covers. What's not to like? :)

Carolyn said I was really "on" tonight, which is nice to hear. :) I certainly had a blast!

More about tonight's episode is here.

 
At January 19, 2008 7:38 PM , Blogger Alex Ferworn said...

Many interesting idea but somewhat misleading. Moore's law may apply to raw computing power but certainly has not applied to any form of machine intelligence. In fact, truth be told, progress has been very slow.

Most of the robots that are approaching deployability for things like military operations, search and rescue, etc. are teleoperated. The brains are back with the human. While there have been great improvements in some basic algorithms (walking, path planning) there have been no every day intelligence breakthrough that I know of. Incremental improvement perhaps but that is also open to debate

Computing fast is not the same as computing intelligent and to imply that they are the same is misleading.

I also quibble with the acceptance of robots. Japan may be a special case. By and large, the acceptance of robot by the elderly is a Japanese phenomena and is not reflected in North America or Europe. It has, I believe, more to do with the particular cultural history and sensibilities of japan. Honestly, my own mother still uses a rotary dial phone and refuses to switch to pressing buttons!

My lab constructed a wheelchair retrofit called NEPWAK (Network-Enable Powered Wheelchair Adaptor Kit) allowing others to control a wheelchair from afar. Our worst critics were the old and disabled who did not like the implication of loss of control. (see: http://www.youtube.com/watch?v=wsE1T9a-3NU)

I also disagree with the idea that robots are not all around us now. Anthropomorphism aside, what do you think you are dealing with when you go to an ATM? You did not give a definition of "robot" but the gist of a common one is "robot == control, activation and sensing in a package". There is a lot of stuff out there that matches this definition including the Hwy407.

Also, why can't robots look like dogs rather than people? Dogs go through all the same doors and stairs that people do.

I have worked extensively with rescue robots being evaluated by (NIST and DHS in the U.S.) for use in urban disasters. By and large, they are impressive for their ability to work where people cannot but they come nowhere near the intelligence of a good dog.

I challenge you to point at one robot that, given 18 months, will exhibit twice the intelligence that it does now. Try and stay away from things like the "Grand Challenge" which change the rules to make things easier--to keep autonomous robots from bursting into flame.

Thanks, good show.

 
At January 19, 2008 8:39 PM , Blogger RobertJSawyer said...

Hi, Alex. Interesting comments. You've got to remember that what you hear in a TV interview are the very first, immediate responses to questions. None of it is scripted or rehearsed, and, in this case, there hadn't been a pre-interview, and the interview is live-to-tape -- no editing. So to be sitting in your easy chair, days later, and drafting and editing a comment at leisure puts you at an unfair advantage when it comes to quibbling. :) As I always say when someone complains that the 30-second summary of any topic given on TV doesn't actually deal with all possible situations, all possible shades of gray, all possible dissenting opinions, all possible interpretations, and so on, can you craft an answer that does all those things AND IS NO LONGER THAN THE ORIGINAL? :)

Also, you have to realize, old boy, that there's only so much time to cover a lot of material on TV; I don't think I give short shrift to any of the issues you raise in my books, but I do think Steve and I did pretty well given the time we had in covering the topics. :)

But I disagree with you about the acceptance of robots by the elderly being a Japanese phenomenon; the studies I was alluding to come from work at Carnegie-Mellon in Pittsburgh with elderly Americans. :)

And you answer your own question about why robots should look like humans -- by citing the negative reaction to your project that was perceived of as a contraption.

And, sure, you can argue about robots being around us all the time, but not many roboticists would agree that an ATM or an automated toll highway [the 407] is a robot -- you define the terms so broadly as to render it useless (a toaster responds to input and manipulates flat physical object, too, just like an ATM manipulating bills, and it's not a robot; if an electronic toll highway is a robot, is one that uses human toll-booth attendants a cyborg?).

:) But I'm glad you enjoyed the show!

 
At January 21, 2008 12:20 PM , Blogger Alex Ferworn said...

Actually, I quite liked your interview and think you did an excellent job. I understand the difficulty in getting an answer out on time in front of a camera and not sounding stupid. You did a very good job but I am questioning your sources. The Robot Institute is only one place and it competes for funding from the U.S. military. Any data that is available should be seen in that light.

Have a look at some of the people who are trying to use robots and their commentary and the reports are shockingly different.

Every year, I go down to Texas and act as an evaluator in a field of robotics called "response". Response robots are intended to assist first responders in urban disasters. The field came into being in 2001 after the World Trade Center disaster. Because first responders at the scene were desperate to try anything to find live humans, they employed various mobile robots that were made available by researchers like Robin Murphy at the U of South Florida.

Because, they provided a degree of "situational awareness" that could not be obtained in other ways they achieved some success and were seen as a potentially useful tool for USAR. The National Institute of Standards and Technology (NIST) through the Department of Homeland Security (DHS) started funding work in the area to standardize what is meant by "response robot".

The belief in 2001 was that rescue robots would be a "snap" and easily made out of various existing research robots and quasi-military robots. Generally speaking, this has not been the case.

Very few signs of intelligence exist in any of these robots and they have a very hard time with the environments that they were designed to tolerate. I have many examples of horrible failures in relatively simple circumstances that were not staged.

As far as a definition of robot is concerned, the research community has a long tradition of looking toward your genre for help. The definition I gave is actually, more or less, one that is being considered as a standard by NIST through the ASTM standardization process. The definition only looks vague if you expect robots to look a certain way and do certain things.

Given my broad useless definition for "robot" is not to your liking, can you provide me one that is?

I have high hopes for anthropomorphic robots that exhibit intelligence in real-time and are durable. However, there are a lot of details before you find a god. Batteries still suck, power cells don't work and are often dangerous, operational life is low, machine vision works in the lab and on highways (kind of). They can't reason there way across a desert yet, they need a multi-billion dollar set of satellites so they only destroy themselves occasionally.

I'm trying dogs. Fast, efficient, smart and they can be augmented with technology.

The Cyborg stuff is interesting to me. Toll booths are a human phenomenon. We need them to interact the way we do because we have limbs and wallets. Why can't I have my 407 robot without limbs? It certainly has a big wallet.

Anyways, keep up the good fight as these ideas are important to discuss in the broader population as they are, and will, keep changing people's lives.

thanks, alex

 

Post a Comment

<< Home