40 years ago I got really excited by artificial intelligence and I thought, incredibly naively, that we could put intelligence and we could put perception and so on into machines but, do you know what, although we failed signally to do that actually, we really did, we learned something. We learned how damn clever the brain is because the brain does things that no way can our technology do at the moment, but having said that, I think now, 40 years later, that computers have developed, we’ve all learned a great deal more in physiology and in psychology, that that dream was actually a good one and I think we should still aim at relating technology to the mind to make mindful machines, to use machines even more for teaching and thinking, tools for thinking and seeing. I think that’s going to be the answer, to live very, very much more closely with intelligent technology, if you like, and I think artificial intelligence really is part of the future. I think it's terribly important, and I think how it should go is that we need to develop computer programmes which don’t simply go through what are called algorithms, little steps of reasoning, but should assess probabilities and I think they should be created so that from past experience it’ll generalise past experience, produce generalisations which are sort of chunks of knowledge really, use these to interpret information, photo-electric cells and so on as in, you know, technologies, and then make the machine think, make it intelligent by, again, relating the available information to its background knowledge to create hypothesis and using what we now call Bayesian strategies which is really the mathematics of inferring with probabilities. And I think instead of little algorithms, this needs to be done, which is a much more ambitious project but I think if we can make that work we’re going to have really intelligent machines and then we can live with them and I think this is how it’s going to go. We’re going to have Bayesian intelligence in machines.
