Be wiser than other people if you can, but do not tell them so.
Lord Chesterfield, Letters to his Son, 1774.
Artificial Intelligence (AI) is the branch of computer science which is attempting to make computers "think".
To many people, AI is highly provocative, because they associate with intelligence other properties which human intelligence bestows, such as:
To these people, intelligence necessarily implies and is implied by these qualities. Clearly computers (at least at the moment) lack all of these qualities. For example, the best computer chess program in the world (which we may release before long) doesn't mind whether it wins or not. In fact, it is just as 'happy' being switched off! It lacks sentience and intentionality and, therefore, by implication intelligence.
One hundred years ago, anyone would have agreed that to play chess (or even calculate square roots) clearly required intelligence. Even today, if a chimpanzee were to play even poor chess, that would be heralded as proof of his intelligence.
So why the difference? The answer is that there are non-intelligent ways of achieving intelligent tasks.
Whatever task is 'defined' as definitely requiring intelligence, once computers can accomplish it and the algorithms are explained, the task is no longer regarded as intelligent "since it can be done by a computer".
The moving goal-posts argument gives what for us is the best definition of AI in the general perception:
What we aim for at Software Scientific is not the holy grail of AI, but the more pragmatic Machine Intelligence:
Just as we use levers and other machines to increase our strength, so we can use computers to increase our effective intelligence. In all our systems, the human remains "in charge", but the computer does the donkey work.