To reiterate, as always: my policy is to print only the parts of letters that raise problems, correct errors, give me an argument, add information, etc.
My policy on the rest of my mail is to enjoy, enjoy, if I can, otherwise ignore, ignore.
Comments of my own are enclosed in parentheses ((like this)).
| March 23, 1971 |
I found your account of the state-of-the-art in robotics most interesting and extremely well done. I particularly appreciate the accurate precis of my own work. There is one minor quibble because I actually tackled the response problem in some detail in ADROIT. The "hunting for grass and twigs" type of programming was not assumed, it was coded to a quite fine level of detail. Aside from that, you are right on. ADROIT was conceived as the first step towards a control program that learned.
...Robotics is in for some hard times right now. Funding is getting tough even for recognized workers. For your information, the Mars Rover project has been cancelled.
Nevertheless, the knowledge is obviously there for the taking. Sometime soon it will be done, and I think that publications like yours may help many people both in and out of the engineering and computing professions to get a better perspective on its significance.
|
Leonard Friedman TRW Redondo Beach, California |
|
|
((That middle paragraph is painful. I had to edit it a bit, but I was anxious to retain the substance of it, because it has such a direct bearing on this month's discussion.
Elsewhere in this issue, I have outlined a strategy for deploying "available boodle". That strategy is based on an assumption that seemed reasonable six months ago: robot research will look after itself, riding the crest of computer research and space research in general. Therefore, anything extra which becomes available (with "artificial reincarnation" in mind) should be used to step up progress in areas that are lagging.
I still think that's a sound strategy if we keep, say, a dozen years in view. There are little setbacks related to the business cycle, and then there are long-range trends. The Mars Rover, for example, is too good to disappear for keeps. By and by things will pick up, and then the work will go forward -- maybe not by the same name, maybe not in one piece, maybe on a more protracted schedule, maybe not tied to the space program, but it'll go forward. The other areas I mentioned, on the other hand, are likely to remain at a disadvantage, because nobody is looking upon them as critical parts of any broader program.
I could be wrong, of course, and if the assumptions prove wrong, the strategy will change. It may be, for example, that the space program, far from being a crest on which to ride, will need some vigorous defending in its own right. It goes without saying that I'm against any cutbacks in robot research, and yes, this publication will do everything it can to make the public aware of the stakes.))
| March 2, 1971 |
Enjoyed your "Word In Edgewise." It, of course, is somewhat out of the realm of comprehension for the average layman. It is my feeling that no machine should replace the human in all fields! To begin with, it is apparent that they are as mistake prone as humans.
Several instances within my experience prove that. Presently I'm in dispute with ___ Insurance Co., which carries my household coverage. Exactly a year ago, it appeared my coverage should be increased, which was done. To this day the "machine" hasn't been able to adjust, and just two weeks ago, a notice of cancellation of policy due to non-payment was received! My local agent is "tearing his hair." Perfect them first!!
|
Mrs. L. F. Singer Great Falls, Montana |
|
|
((That last is from my aunt, who has little to do with computers -- though more, it would appear, than she chooses. I suspect that she speaks for a lot more of my regular readers than have ventured to say so. Irritation with computers is rather widespread these days, and like flu, it cuts right across the political spectrum.
I could answer on several levels. Perhaps the simplest answer is that we aren't talking about the same "machines." I wonder how many outsiders know that when the Patent Office finally decided to issue patents on computer programs, the consideration that moved them was this: the same piece of hardware operating under different programs amounts to different machines.
A computer which is programmed to make up bills and keep track of payments isn't what you'd call an "artificial intelligence." Nobody designed it to be clever -- just efficient and cheap. It earns its way by doing a few simple routines over and over again, mindlessly. In other words, it works on an assembly line.
When human beings work on an assembly line, they are sometimes said to be "dehumanized." I won't make myself ridiculous by suggestion that a computer in the same situation is "de-computerized." I will say, though, that machines are no better able than men to show off their full potential, when they are programmed to be mindless cogs in a system of mass production. To be sure, they lend themselves more readily to being machines -- but the question is, what more can we make them?
Beyond that, I did state that the hardware itself is inadequate at present, and we can't even specify very well what kinds of improvement we'd like to see. Believe me, I don't propose to transplant human beings into any lower form of life.
One other comment: in some ways, machines are too perfect. Maybe we're more creative than machines precisely because we're "imperfect" in certain ways. Human beings are misinformed, illogical, emotional, etc. -- so we try things that make no sense to try, and we blunder into great discoveries.
Anyhow, to err is human, and I don't think we'd feel comfortable being transplanted into something inhuman. What I'd like to see is not artificial divinity, but an artificial intelligence that is perfect and imperfect in pretty much the same ways as human beings.))
|
|
A Word In Edgewise #8, June, 1971: |
| The Isabella Question: Should the Queen Pawn Her Jewels? |
 |
 |