A fundamental problem in artificial intelligence is that nobody really knows
what intelligence is. The problem is especially acute when we need to consider
artificial systems which are significantly different to humans. In this paper
we approach this problem in the following way: We ta