The Trouble With A.I. - Part II
Not Altering Public Perception
Although the public perception of A.I. is based on science fiction rather than reality on one level it is correct - machines WILL take over the world. It's no good trying to convince people otherwise when they can perceive it happening (however slowly) in their everyday lives. What needs to be changed is peoples idea of the rate and degree of this process. A.I. isn't suddenly going to appear and within minutes bring about the apocalypse - and we won't suddenly see blade-runner style replicants drinking coffee in starbucks. Instead A.I. will start out stupid and over time get smarter. Initially it will be used in very specific domains and only with time will any level of 'general' intelligence be gained. All this will happen at a pace most people won't even notice.
What is A.I.?
The issue that always seems to crop up in conversations I have with people (both A.I. aware and not) is "What is (Artificial) Intelligence?" - and it's a conversation stopper. Its well known to any undergrad in the field that you can grind a seminar class to a halt by dropping this question into the discussion. Its equivalent to using the line "Ah - but what is reality?" in any other conversation. The strange thing is that I've not actually met anyone working in the field that is particularly concerned about the answer to the question. And why aren't they? Because knowing the answer won't assist them in their work. The fact is that "Artificial Intelligence" is a rather outdated term for the range of work that goes under that label - there just doesn't seem to be a better one around at the moment.
The "What is intelligence" question isn't really a proper question - or, to be more accurate, hasn't got a proper answer. The best answer I can give you is "Intelligence is something that some things have more of than other things". To illustrate my point I'll attempt to categorise 5 different things into the groups 'intelligent' and 'not intelligent'. My chosen 'things' will be :-
Hmm...lets see. Well I'm intelligent - that's obvious (and strangely - the person doing the defining always thinks this is the case). Now my cat - well...I suppose if I wasn't around it could find its own food, it does manage to stay out of trouble most of the time and it does seem to have some concept of cause/effect and the difference between things in the world...so I think I'll consider it to be intelligent. Now the dolphin - I guess most of the things I mentioned about my cat would hold true for a dolphin to - so I'll put that on the intelligent side as well. Ah...now we come to the ant...well...ants don't seem to do very much...they just crawl about - its not really intelligent as such...but I'm not sure - it does stuff that's pretty similar to a dolphin and a cat - its just smaller...maybe I should reconsider putting cat and dolphin in the 'intelligent' group after all.
And herein lies the problem. Intelligence is not a fixed criterion - it's a sliding scale - there's not some 'magic point' after which something can be called intelligent - its just something that some things have more of than other things. By the way - the brick isn't intelligent (and never will be) - but that's just my opinion...the brick might see it differently.
Creepy A.I.
A.I. at the current time is, at least in the eyes of the general population, seen as a 'controlling' technology - this needs to be challenged. Many of the best forms (or applications) of A.I. are 'assistive' rather than 'controlling' in nature - i.e. A.I. that allows users to make better decisions rather than making the decisions for them. I see 'Assistive Technology' as an area where A.I. can beat more traditional methods hands-down.
Take, for example, the predictive text function on a mobile phone - is it intelligent? Most people would say no - but it can do something that no animal or even a young child can do - i.e. give suggestions for possible words based on a partial set of characters. Although the text predictor doesn't alter its own code or adapt very much (if at all) it does perform searches in the very specific domain of 'word space' very well. If we consider the previously discussed 'sliding scale' of intelligence it might rank somewhere near the bottom...but it is on there. If that is the case then A.I. is already being used by millions of people all over the world - and they didn't even notice.
If other forms of more intelligent A.I. could be integrated into peoples lives - as an assistive technology - then, like the computer, A.I. will begin to creep up on the general population without them fearing some unlikely apocalypse.
Although the public perception of A.I. is based on science fiction rather than reality on one level it is correct - machines WILL take over the world. It's no good trying to convince people otherwise when they can perceive it happening (however slowly) in their everyday lives. What needs to be changed is peoples idea of the rate and degree of this process. A.I. isn't suddenly going to appear and within minutes bring about the apocalypse - and we won't suddenly see blade-runner style replicants drinking coffee in starbucks. Instead A.I. will start out stupid and over time get smarter. Initially it will be used in very specific domains and only with time will any level of 'general' intelligence be gained. All this will happen at a pace most people won't even notice.
What is A.I.?
The issue that always seems to crop up in conversations I have with people (both A.I. aware and not) is "What is (Artificial) Intelligence?" - and it's a conversation stopper. Its well known to any undergrad in the field that you can grind a seminar class to a halt by dropping this question into the discussion. Its equivalent to using the line "Ah - but what is reality?" in any other conversation. The strange thing is that I've not actually met anyone working in the field that is particularly concerned about the answer to the question. And why aren't they? Because knowing the answer won't assist them in their work. The fact is that "Artificial Intelligence" is a rather outdated term for the range of work that goes under that label - there just doesn't seem to be a better one around at the moment.
The "What is intelligence" question isn't really a proper question - or, to be more accurate, hasn't got a proper answer. The best answer I can give you is "Intelligence is something that some things have more of than other things". To illustrate my point I'll attempt to categorise 5 different things into the groups 'intelligent' and 'not intelligent'. My chosen 'things' will be :-
- Me
- My Cat
- A Dolphin
- An Ant
- A Brick
Hmm...lets see. Well I'm intelligent - that's obvious (and strangely - the person doing the defining always thinks this is the case). Now my cat - well...I suppose if I wasn't around it could find its own food, it does manage to stay out of trouble most of the time and it does seem to have some concept of cause/effect and the difference between things in the world...so I think I'll consider it to be intelligent. Now the dolphin - I guess most of the things I mentioned about my cat would hold true for a dolphin to - so I'll put that on the intelligent side as well. Ah...now we come to the ant...well...ants don't seem to do very much...they just crawl about - its not really intelligent as such...but I'm not sure - it does stuff that's pretty similar to a dolphin and a cat - its just smaller...maybe I should reconsider putting cat and dolphin in the 'intelligent' group after all.
And herein lies the problem. Intelligence is not a fixed criterion - it's a sliding scale - there's not some 'magic point' after which something can be called intelligent - its just something that some things have more of than other things. By the way - the brick isn't intelligent (and never will be) - but that's just my opinion...the brick might see it differently.
Creepy A.I.
A.I. at the current time is, at least in the eyes of the general population, seen as a 'controlling' technology - this needs to be challenged. Many of the best forms (or applications) of A.I. are 'assistive' rather than 'controlling' in nature - i.e. A.I. that allows users to make better decisions rather than making the decisions for them. I see 'Assistive Technology' as an area where A.I. can beat more traditional methods hands-down.
Take, for example, the predictive text function on a mobile phone - is it intelligent? Most people would say no - but it can do something that no animal or even a young child can do - i.e. give suggestions for possible words based on a partial set of characters. Although the text predictor doesn't alter its own code or adapt very much (if at all) it does perform searches in the very specific domain of 'word space' very well. If we consider the previously discussed 'sliding scale' of intelligence it might rank somewhere near the bottom...but it is on there. If that is the case then A.I. is already being used by millions of people all over the world - and they didn't even notice.
If other forms of more intelligent A.I. could be integrated into peoples lives - as an assistive technology - then, like the computer, A.I. will begin to creep up on the general population without them fearing some unlikely apocalypse.