hakanson@orstcs.UUCP (12/01/83)
#R:ucf-cs:-109500:orstcs:11600021:000:777
orstcs!hakanson Nov 29 20:58:00 1983
The other disadvantage of "The Library" is that dependence on it can lead
to the kind of Big Brother censorship evidenced in Startide Rising, when
they keep trying to find info about the mummy they retrieved, and they
keep getting "I never heard of it." Obviously (as stated in the book),
the progenitor-races that gave them the thing fiddled with the circuits
so it wouldn't help them too much.
As a counter-future-history, notice that the Terrans don't rely totally
on the Library -- in fact, they make a point of diverging from known
technology whenever possible, sometimes just in order to be perverse.
The defense against such stagnation/dependency/censorship? -- Skepticism.
Marion Hakanson CSnet: hakanson@oregon-state
UUCP : {hp-pcd,teklabs}!orstcs!hakansonandree@uokvax.UUCP (12/02/83)
#R:ucf-cs:-109500:uokvax:5400012:000:1679 uokvax!andree Nov 30 20:35:00 1983 I think you just managed to violate the canonical behavior of futurologists (optimistic in the short run, pessimistic in the long run). There are ALREADY medical & r-d expert systems out there. PUFF (pulmonary disease diagnosis & treatment), CASNET (ditto glaucoma), MYCIN (ditto infectious diseases), CADUCEUS (internal medicine), SOPHIE (EE instructor), and R1 (configuring DEC hardware!) can be added to PROSPECTOR (geology) and DENDRAL (chemical structures). Most of these are experimental, some (most notable R1 and PUFF) are in regular use outside the CS lab. I think your 2000 deadline (expert systems in regular use) will be here by 1990. Your other deadlines seem to depend on having machines that we could really call `intelligent' - that could deal with unexpected situations from the entire real world. I have a feeling that that is still a long way off. Expert systems are very good - in a very small, specialized area. The reason they look so good is that they deal with a small (85% of internal medicine small!?! GACK!) database when compared with the real world. Your point of adding good, general purpose robots to The Library is good, though. It would only be logical for fact X in the library to carry a tag about how to verify fact X. So having a robot run the verification for you would be logical (historical facts excepted, of course). Of course, this could make experimentation on your own doubly gauche: first because you didn't look it up in The Library, and then again because you actually DESIGNED AN EXPERIMENT! GASP! <mike P.S. You did make me wander - how would an artificial intelligence fit into Brin's Universe? Property? Uplifted race?
giles@ucf-cs.UUCP (Bruce Giles) (12/03/83)
My whole point was that, like it or not, we will soon have the conterpart of a "Library" ourselves. In fact, we have the beginnings of one now. When was the last time you solved a non-trivial integeral without using a reference book? I know that I normally use the CRC tables, or the one by Abramowitz and Steigum (sp?). How long will it be before such systems go on-line? Before entire classes of problems are solvable using "expert knowledge" systems. I have heard interesting things about general relativity and computers; how long until *only* computers can handle the actual mathematics in such areas? I agree that we should never take such answers for granted, but what about when we do not *have* the ability to check the results of a Library without using a Library? I presented my scenario earlier to point out the fact that we will be in precisely that position in the near future. (ok -- I admit it, I can think of millenia as short periods of time. :-) ) Finally, perhaps we should split this discussion into two headers: one based on the universe of *Startide Rising*, the other based on our own based on AI research. ave discordia Bruce Giles --------------------------------------------- UUCP: decvax!ucf-cs!giles cs-net: giles@ucf ARPA: giles.ucf-cs@Rand-Relay Snail: University of Central Florida Dept of Math, POB 26000 Orlando Fl 32816 ---------------------------------------------
andree@uokvax.UUCP (12/16/83)
#R:wu1:-19900:uokvax:5400013:000:816 uokvax!andree Dec 6 11:52:00 1983 /***** uokvax:net.sf-lovers / wu1!rf / 5:22 am Dec 2, 1983 */ It may be that, at some level, *thinking* is a process which itself is limited. Even if not, wise sentients might not wish to construct sentients who could outthink them. So perhaps the Galactics never got about a millenium ahead of us. Randolph Fritz /* ---------- */ This brings up a possibility raised in other stories. The contention is that by reaching some specific level of civilization, the traits selected for are changed - in such a way that the thinking abilities (dare I say intelligence) stops going up. Given this hypothesis, the Galactics CAN'T be more than about a millenium ahead of us. Of course, writing about a race that is actually far ahead of us in mental abilities is an interesting challange in and of itself. <mike