Marvin Minsky is here critical of many current researchers in artificial intelligence researchers who he feels have gotten bogged down in theories of machine learning. He sees this as a crisis point in a time of an aging population that he feels will need help in performing many tasks.
"We have a computer program that can beat a world chess champion, but we don't have one that can reach for an umbrella on a rainy day, or put a pillow in a pillow case." For a machine to have common sense, it must know 50 million such things, and like a human, activate different kinds of expertise in different realms of thought, says Minsky.
The machine he envisions will have a very high-level, rule-based system for recognizing certain kinds of problems like humans do.
Then he, in his good scientist way, classifies things - like the parts of the brain he calls â€œcriticsâ€ that are selected for a particular situation, while other critics turn off. His machine reasoning architecture has 6 levels of thinking that attempt to emulate the different kinds of human reasoning such as learned reactions, deliberative thinking, reflective thinking (sometimes several simultaneously).
Complication is necessary: with at least 400 different areas of the brain operating, â€œif a theory tries to explain everything by just 20 principles, it's doing something wrong.
I actually take some hope in his despair about AI because it tells me that scientists are still trying to understand some of the amazing but "simple" things that we humans do. What he is really proposing is a kind of AI that might eventually result in a â€œreally resourceful, clever thinking machine...with knowledge about how to do things,â€ and which â€œcan do the broad range of things children can do.