Comic Discussion > QUESTIONABLE CONTENT
WCDT: 18-22 Oct 2010 (1776-1780)
Skewbrow:
--- Quote from: Olymander on 23 Oct 2010, 08:58 ---In either case, though, I would think that something "smarter" is more likely to arise by accident, or randomness, than by design.
--- End quote ---
Agreed.
Is it cold in here?:
--- Quote from: Olymander on 23 Oct 2010, 08:58 ---How do we define something as "smarter than ourselves", though?
--- End quote ---
Insightful question. For purposes of Singularity theology, a system smarter than us is one that is better than us at designing even more sophisticated systems.
I'm saying "system" rather than "computer" because it takes whole ecosystems to create something like a Sun server or a Cray.
Near Lurker:
--- Quote from: Skewbrow on 23 Oct 2010, 11:03 ---Our brain cannot "approximate infinitely many algorithms", because it only has had a finite amount of time to learn, and only has a finite pool of things to try.
--- End quote ---
The thing is, though, that the problems to be solved don't all have the same likelihood of coming up (after all, the first thing anyone thinks seeing these proofs or paradoxes is "...but that's a pathological case"), so it's possible to reach a finite fraction of relevance, even quite a large one, in finite time. And, to the point of the argument, the human brain almost certainly can't solve any problem, or at least it hasn't been proven to be able to, and most of what we know about neuroscience suggests it no more can than a computer.
--- Quote from: Skewbrow on 23 Oct 2010, 11:03 ---However, the "haphazard" part is more promising. Some of the more curious stuff revolves around concepts like "genetic algorithms". Even they are still run on Turing machines and are thus somewhat limited in their capabilities. Also we then lose the blinding speed that we today associate with computers.
--- End quote ---
The Turing machine is a thought experiment-cum-mathematical construct modeling the evaluation of algorithms, one that is in some ways less limited than a modern computer and in some ways more. It's not a smart-sounding word for "computer."
Skewbrow:
--- Quote from: Near Lurker on 23 Oct 2010, 12:10 ---The thing is, though, that the problems to be solved don't all have the same likelihood of coming up (after all, the first thing anyone thinks seeing these proofs or paradoxes is "...but that's a pathological case"), so it's possible to reach a finite fraction of relevance, even quite a large one, in finite time.
--- End quote ---
This is certainly true for many problems (but IIRC not all of them). This I won't challenge. For example the tiling problem is easily solvable by humans in many cases of practical interest (this is a somewhat silly argument, because in this case by definition "practical interest" implies "solvable by humans"). You need to go to something really wacky for the computer to balk. An interested reader may use "Penrose tiling" as a buzzword for a search engine, though the relevance of the Penrose construction to the tiling problem is elsewhere.
--- Quote from: Near Lurker on 23 Oct 2010, 12:10 ---The Turing machine is a thought experiment-cum-mathematical construct modeling the evaluation of algorithms, one that is in some ways less limited than a modern computer and in some ways more. It's not a smart-sounding word for "computer."
--- End quote ---
:-) Mathematicians/theoretical computer scientists use the word "Turing machine" because it has a precise definition, whereas the word "computer" has not. This is obviously necessary in order to say something precise about the capabilities of a device. I used it here so that eventual interested folks (if any?) can use it as a buzzword when searching for more information (and in order not to say something untrue). However, I'm under the impression that all current computers could (in theory) be emulated by a Turing machine, so the concept is not without practical merit. For any real work a Turing machine would be awfully clumsy, because the computers have useful add-ons like user interfaces and dedicated circuitry for frequently occuring tasks, but the range of tractable problems is the same (if the computer had access to unlimited amounts of memory).
Carl-E:
--- Quote from: Skewbrow on 23 Oct 2010, 01:34 ---My math PhD is in rather different area...
--- End quote ---
Wait, how many mathemeticians are on this board?
And Skewbrow, what's your field? Mine's Knot Theory (classical dimensions), Indiana University, '95.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version