Posted by
michael
on from the dreaming-the-future dept.
Anonymous Coward writes: "Ray Kurzweil and other digerati discuss when popular sci-fi concepts will manifest in the real world. See part I or
part II."
"Futurists"
by
SeanAhern
·
· Score: 3, Interesting
It has been said that so-called "futurists" oversell the short term, and undersell the long term.
Machine translation? You gotta be kidding!
by
epsalon
·
· Score: 4, Interesting
Machine translation in 0-30 years?! As a person involved in these topics, I can say that 30 years ago people thought this could be solved in 30 years. We are today almost as far away as we were 30 years ago, and I think that there's no way of this being a realitiy in less than 100 years.
To do correct machine translation you have to fully model the world and knowledge. Translation (for humans) is a tedious job, requiring a lot of research and artistic-like choice of words.
I think that we will sooner have machines writing their own novles than full machine transtaltion. The problem is just too hard.
The problem with the Turing test
by
evilviper
·
· Score: 3, Interesting
The biggest problem with the Turing test is that it is completely subjective. The smarter of a person you are, the smarter the computer will have to be to give an accurate response. Obviously that trait is not one that reflects intelligence.
Get someone dumb enough and they'll chat with ELIZA for hours at a time.
Using the brain to store digital information:
The problem is less one of interface than it is one of reprogramming neurons. While this might technically be possible, is there going to be any sort of information density advantage? Human memory has some really nice lossy compression, but that would make it a bad way to store digital data.
Computers "understanding" and "speaking" human language:
I think the only thing we've really learned in the last 30 years is that the problem is a lot harder than we thought it was 30 years ago. There are a multitude of problems, from simple parsing to having a large enough database to understand context. That, and we really don't know what problem we are solving. A speech interface to a database would seem to be to be a useful tool - "what is the weather going to be like today?" opens up the appropriate web page. "Find me a good price on a 1997 Honda Accord" hits the search engines, finds a few dealers in my area, and gets me some pages to view. We don't even have anything this sophisticated without the voice interface. (Speech-to-text + text-to-speech + Google) is not tons better than Google. Yet, we expect a program with the depth of knowledge and subtlety of reasoning that a human posesses. My own version of the Turing Test, "I'll believe it when I see it," suggests to me the system that can pass the Turing Test is a LONG way off.
Software as a weapon:
OK, ID was a poor example - I know I'm 1337 enough to reverse engineer alien technology in a matter of minutes and write a virus using a Mac, but that guy? But really, software as an weapon is only useful against those who use software, and only when that software is of critical importance. Even North Americans aren't THAT reliant on the 'net, although it might be wise to take precautions before we wire all of our brains together...
Slashdot Mirror
It has been said that so-called "futurists" oversell the short term, and undersell the long term.
Machine translation in 0-30 years?! As a person involved in these topics, I can say that 30 years ago people thought this could be solved in 30 years. We are today almost as far away as we were 30 years ago, and I think that there's no way of this being a realitiy in less than 100 years.
To do correct machine translation you have to fully model the world and knowledge. Translation (for humans) is a tedious job, requiring a lot of research and artistic-like choice of words.
I think that we will sooner have machines writing their own novles than full machine transtaltion. The problem is just too hard.
Make even shorter URLs - 8LN.org
The biggest problem with the Turing test is that it is completely subjective. The smarter of a person you are, the smarter the computer will have to be to give an accurate response. Obviously that trait is not one that reflects intelligence.
Get someone dumb enough and they'll chat with ELIZA for hours at a time.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
The geek says it will all happen, it's just a matter of time.
Using the brain to store digital information:
The problem is less one of interface than it is one of reprogramming neurons. While this might technically be possible, is there going to be any sort of information density advantage? Human memory has some really nice lossy compression, but that would make it a bad way to store digital data.
Computers "understanding" and "speaking" human language:
I think the only thing we've really learned in the last 30 years is that the problem is a lot harder than we thought it was 30 years ago. There are a multitude of problems, from simple parsing to having a large enough database to understand context. That, and we really don't know what problem we are solving. A speech interface to a database would seem to be to be a useful tool - "what is the weather going to be like today?" opens up the appropriate web page. "Find me a good price on a 1997 Honda Accord" hits the search engines, finds a few dealers in my area, and gets me some pages to view. We don't even have anything this sophisticated without the voice interface. (Speech-to-text + text-to-speech + Google) is not tons better than Google. Yet, we expect a program with the depth of knowledge and subtlety of reasoning that a human posesses. My own version of the Turing Test, "I'll believe it when I see it," suggests to me the system that can pass the Turing Test is a LONG way off.
Software as a weapon:
OK, ID was a poor example - I know I'm 1337 enough to reverse engineer alien technology in a matter of minutes and write a virus using a Mac, but that guy? But really, software as an weapon is only useful against those who use software, and only when that software is of critical importance. Even North Americans aren't THAT reliant on the 'net, although it might be wise to take precautions before we wire all of our brains together...
sig fault