MadSci Network: Computer Science
Query:

Re: Do you think true artificial intelligence will ever be achieved?

Area: Computer Science
Posted By: Kent Dirckx, Grad Student, Child Development, University of Minnesota
Date: Wed Jul 23 08:15:54 1997
Area of science: Computer Science
ID: 865569443.Cs
Message:

This is a tough question to answer, but let me share a few thoughts on the topic:

The first question is "what is 'true' artificial intelligence?" We all think that we'll know it when we see it, but we really need some form of test to determine when we have reached artificial intelligence. Alan Turing, one of the fathers of artificial intelligence research, prposed the following test, now called the "Turing test" and is a classic in AI:

Imagine that you are sitting in front of a computer and you are talking to two people online. However, one of the "persons" is actually a computer that is responding to your questions. Your job is to tell which "person" is actually the computer and it is the job of both the person and the computer to convince you that they are human. You can say or ask anything to both, but the computer will not necessarily answer truthfully. If the computer fools you into thinking that it is the human, it has passed the Turing Test and is considered intelligent by this measure.

Now we can ask "will we achieve a system capable of passing the Turing Test?" I think so, science is filled with people doing what others say cannot be done. Artificial Intelligence is complex and will require a lot of knowledge that we don't have yet, but there is no reason why it cannot be accomplished.

My prediction is that we will change the way we think about intelligence to allow for intelligence that is not human-like intelligence. The pursuit of passing the Turing Test involves giving a computer not only human strengths (language understanding, creative thought) but also human weaknesses (slow response, occassionally wrong answers.) Giving computers human weaknesses seems to be a large sacrifice to create an artificially intelligent machine. Even now, the popular neural network models allow for a system to learn incorrectly and make wrong answers. Computers were designed to be fast number-crunchers and logic machines while humans are creatures of creativity and intuition. The two ways of processing information are very different and complement each other well. Hopefully, we can define intelligence in such a way that a computer does not have to sacrifice its strengths in order to qualify.


Current Queue | Current Queue for Computer Science | Computer Science archives

Try the links in the MadSci Library for more information on Computer Science.



MadSci Home | Information | Search | Random Knowledge Generator | MadSci Archives | Mad Library | MAD Labs | MAD FAQs | Ask a ? | Join Us! | Help Support MadSci


MadSci Network
© 1997, Washington University Medical School
webadmin@www.madsci.org