Previous Entry Share Next Entry
Part of the Essay I got 80% on...
2012
unknownj
I had to analyze 15 different statements about AI, with respect to four dimensions of classification. I then had to write my own definition of AI, and classify it in a similar fashion. Here's what I came up with:


"The construction of a machine such that its operation is indistinguishable from so-called 'naturally occurring intelligence'. That is to say, to create a machine that is as 'good' as any living creature at being 'alive'." - James Webley

My own personal definition of Artificial Intelligence is somewhat unrealistic, to say the least. Such a machine will certainly not be created in my lifetime, or indeed the next few hundred years (although such predictive claims generally end up being proven very wrong). However, to my mind, this is what artificial intelligence is striving toward ultimately, and it is (to me, at least) the very definition of what artificial intelligence really ought to be. Primarily, thought processes and reasoning would be paramount to its operation, but this is not to say that behaviour and performance would be overlooked. This is, to my mind, a universal definition of artificial intelligence, which encompasses both aspects of the first dimension of classification.

Success would be measured in terms of human performance rather than performance against some logical rational ideal intelligence - to my mind, if billions of years of evolution is incapable of creating pure ideal intelligence, then perhaps that's because it is non-optimal. As far as humanity is concerned, the end result of evolution thus far is us - striving for anything else would be contrary to what nature has led us to believe is 'the best'.

At first, this definition is concerned with theoretical results. After all, the most crucial step (and therefore the first to concentrate on) would be the thought processes of the machine. However, ultimately it would be useless without any practical applications - there is only so far you can stretch the usefulness of theoretical results. Additionally, the realist in me knows that ultimately research is funded by those people whose interests are best served by the end result - without practical applications, it is a lot harder to secure this financial backing.

Due to the nature of my definition, consciousness is a definite requirement - how else can one truly create something indistinguishable in its thought from humans? How one determines whether the machine is conscious is a slightly harder task, but as stated in "Computing Machinery and Intelligence" by Alan Turing, if a machine gives the impression of having consciousness, it would be a very solipsist point of view to continue to refute the presence of consciousness.

I prefer this definition to the others simply because it is the most romantic - it describes the eventuality that countless science fiction writers have described. It gives a clear final destination for artificial intelligence, but does not preclude the possibility that all other definitions can be explored on the way.


I was then asked if the fact that there are some things that it can be mathematically proven that computers will never be able to do means that Computers cannot be intelligent. I answered thus:


The fact that there are problems that a computer cannot solve is irrelevant to the feasibility of artificial intelligence. Consider humans - there are many problems that we are unable to solve or even contemplate. It would take a lot of convincing in order to make the entire world believe that the fact that we have limitations precludes the possibility that we are intelligent. It seems likely to me that there are problems in the world that we cannot even begin to understand, and thus are unaware of. Because we cannot solve these, it does not make us unintelligent.

Ultimately, of course, if we are talking about boundless artificial intelligence, then yes, this does make it impossible. But Artificial Intelligence is not defined as a technology that can do anything, and consider any problem. In fact, it is not technically defined as being anything static. It is impossible to pin down and call impossible something that you cannot even define satisfactorily to everybody.


Well I'm impressed, anyway... The finished product was a 3000 word essay, which I wrote in an hour and fourty minutes - not bad going, really... :o)

?

Log in

No account? Create an account