That was a great summation Kees. We are attempting to teach a computer (or robot) to "think" and develop independent solutions to new problems. This solution is years in the future. Like any other long journey, we have taken the first step and get closer every day.
Jeff Hawkins has some interesting ideas, and has done some good research and development work, but I have some problems with his assumption that bayesian networks and so on, actually capture the base case of brain function. My own work is not nearly as well accepted however, nor as well researched and developed so at the present Similarity Selection is no competition for his company.