The most common definition of artificial intelligence (AI) is that it is intelligence as exhibited by machines. Well, that’s fine as far as it goes, but it leaves two questions:
* What do we mean by intelligence?
* What constitutes a machine?
Attempts to define intelligence lead to answers like:
* Intelligence has been defined as things like having the capacity for: logic, problem solving, creative work, learning etc. The list goes on.
To take just one of those words (learning), a sparrow can learn that if it comes to the same place every day it will find breadcrumbs there. Does that mean that the sparrow is intelligent? Some will say yes; some will say no; some will say it depends on how you define intelligence, which takes the question right back to the start. And if examination of that word is followed up by, say, “creative work,” it leads inevitably to the next question: “If a sparrow can learn how to get a feed of breadcrumbs, does it also have the ability to write a symphony?”
The answer to that question, so far as anyone alive today is aware, is “No.” It seems, then, that there are gradations of intelligence from the most basic to the most advanced.
Trying to define a machine does not advance things, because the classically accepted definition of machine is the one used in physics: a machine is a tool containing at least one part that uses energy to perform an intended action. That unmodified definition is of so little practical use in the discussion of artificial intelligence that this website will discard it; artificial intelligence here will be regarded as the capacity to learn, to solve problems and to make and/or decisions of some complexity.
As will become apparent, even that simple definition will cause problems.