The challenge of artificial intelligence part 1

Editor's Notes

• #14 Let’s round up our view of computer and AI history. We have made enormous progress with the digital algorithmic computer but it has now matured and doesn’t appear to be going anywhere from here. We have lots of questions that we want answered which current computers will not be able to answer. AI has had a very chequered history. In its first 30 years it promised much and delivered little. And paid the penalty – money was withdrawn. Yet more recently it has surprised us by actually creating useful things, with the promise of more to come. This is both exhilarating as well as frightening. There are two approaches appearing. First is what is called Strong AI. It is heavily reductionistic. That is, intelligence is simply the operation of principles which are founded on more basic principles and so on all the way down. We can decompose a human being and discover the principles that make them work, then replicate. Second is Weak AI. It is dominated by people who want to solve human questions. Their solutions replicate intelligence in one way or another but it is mimicking something that is known but not really understood. AI is here to stay. Our focus will be on Strong AI. This image is beautiful. From the 1927 movie Metropolis (Fritz Lang). It will continue to stretch our understanding of problems and how we analyse them, how we discover information and how we physically perform them. There are dangers. Unemployment being most notable. But that is something that can be managed rather than avoiding progress. But the ongoing pursuit of understanding and replicating humanity is still at the starting blocks. Not much progress at all, despite lots of predictions and noise. This is given the title Strong AI. But replicating a human being, well that requires us to answer a number of questions about what intelligence is and what being a human being is. This is the subject of our second half.