My ears are pleasantly pink with the reaction to issue #5. In fact, I'm sorry now that I so efficiently mimeographed all but the covers and the letter columns for four issues in a row. I could easily dispose of half again as many copies as I ran off, with some assurance that they'd be read.
I should add that while it's my policy to print only the parts of letters that raise problems or add information, I do appreciate the mail I don't print. Encouraging letters such as I received this time from Carl Hewitt of M.I.T. and Deena Koniver of N.I.H. are good for an afternoon's cheer each, and maybe a more constructive attitude the morning after.
So here we go with readers. Comments of my own are enclosed in double parentheses ((like this)).
| Feb. 12, 1971 |
I agree with nearly everything you said and am working on a book which will say the same things plus a great deal more.
One point where we differ is the transplant concept. I have a clear concept of a machine functioning like a man. But there is some difference between (1) make machine like man, then kill the man, (2) have the man climb into the machine. But I am not sure just what this difference is, or how you test for it. I seem to recall a S.F. story wherein if you flunk your Turing Test you get shot.
But there is an easier way. I predict that before the end of the century you will be able to get a shot (tailor-made to suit the individual) to restore cell chemistry to approximately that of a young adult. This is already being done in a modest way.
|
John K. Dixon National Institutes of Health Bethesda, Md. |
P.S. You will enjoy Artificial Intelligence: The Heuristic Programming Approach, by my boss, Jim Slagle. (McGraw-Hill, 1971.)
Another thing, not all business executives are unlettered.
|
|
((Here's a test I've been thinking of. At some point in the "transplant" process, the man must be aware of both flesh and robot as equally part of himself. The (natural) man looks up from the operating table and says, "Peter." The robot says, "Piper." The man says "picked," Like that, only rapidly and smoothly, the old half and the new half bandy some tongue-twister between them. The examiner then has the pair answer a number of unrehearsed questions in this mode, until it is convincingly demonstrated that one consciousness is experiencing both embodiments.
I don't know what comes next: maybe the flesh-and-bonely part says, "O.K., you can turn this part off, now." Ideally, one might prefer to maintain dual control (by wireless?) until the doomed half expires in its natural time. But after all, the consciousness is the thing -- we amputate limbs, don't we? What's crucial is that there be a convincing sense of continuity, attested to by the prospective expiree as well as the continuator.
Speaking of good books, I regret my failure anywhere to mention Semantic Information Processing, edited by Marvin Minsky, The MIT Press, 1968. I had it on my bookshelf all the while that I was writing this series -- and had to choose between reading or writing.
For the half or more of my readers who aren't connected with computing, I should point out that James Slagle was an early arrival in the field of artificial intelligence, with a program that did freshman calculus problems. Out of that and similar efforts came a branch of programming called "heuristics," and if my information is correct, Slagle's project at the National Institutes of Health is a National Heuristics Laboratory.))
| Feb. 12, 1971 |
My great worry is not for the individual life, nor even for the life of the whole human race, but for life and accomplishment itself. Assuming the "BIG BANG" or "PULSATING" or "STEADY-STATE-WITH-UNDERTOW-TOO-BIG-TO-SWIM-AGAINST" theory of the Universe is valid, it does no good to invent an "immortal man."
When the famous 3rd law of thermodynamics takes over in the end, we'll all starve to death (or go into hibernation to last out the PULSE and be gradually eroded away). When the Universe reaches near-equilibrium, there will be no way for man nor machine to survive.
|
Robert Maas Stanford A.I. Project Stanford, California |
|
|
((All right, "immortality" may be an exaggeration. Even if my hopes are realized, most of us will meet some accidental demise long before the Universe throbs. I confess to more desperate concerns: as matters stand, we're all condemned to die within the traditional span. Two lifetimes in place of one would be immortality enough to start with.
Your concern for "life and accomplishment itself" is magnanimous, but are you sure it's real? What can you do about it? On that basis, is any future whatsoever better or worse than another, as long as something wiggly finds its way into the library? It looks to me like one of several dodges which people use to avert their eyes from what's really bothering them -- the fell decree of mortal doom -- by identifying with something that will outlast them.
Rationalizations are acceptable when they're needed to preserve your sanity -- but I'm suggesting we can beat the rap.))
|
|
A Word In Edgewise #7, April, 1971: |
| Pattern Recognizer Meets Theorem Prover: A Robot Gets its Head Together. |
 |
 |