|
|
|
The notion of an ethical machine can be interpreted in more than one way. Perhaps the most important interpretation is a machine that can generalise from existing literature to infer one or more consistent ethical systems and can work out their consequences. An ultra-intelligent machine should be able to do this, and that is one reason for not fearing it.
Introduction
There is fear that 'the machine will become the master', especially compounded by the possibility that the machine will go wrong. There is, for example, a play by E. M. Foster based on this theme. Again, Lewis Thomas (1980) has asserted that the concept of artificial intelligence is depressing and maybe even evil. Yet we are already controlled by machines - party political machines.
The urgent drives out the important, so there is not very much written about ethical machines; Isaac Asimov wrote well about some aspects of them in his book I Robot(1950). Many are familiar with his 'Three Laws of Robotics' without having read his book. The three laws are:
"1. A robot may not injure a human being, or, through inaction, allow a human being
to come to harm.
2.A robot must obey the orders given it by human beings except where such orders
would conflict with the First Law.
3.A robot must protect its own existence as long as such protection does not conflict
with the First or Second Law."
Originally, I thought the three laws were mutually incompatible because they are not quantitative enough, but I found that Asimov had not by any means overlooked the quantitative aspect.
In one chapter of the book a robot on another planet refuses to believe that men, inferior as they are, can construct robots, and it also does not believe that Earth exists. Nevertheless the robot has religious reasons for keeping certain pointer readings within certain ranges, and it thus saves Earth from destruction. Thus the robot does not violate the first law after all. I was unconvinced by this idea, but it does suggest the possibility of a robot's being largely controlled by its 'unconscious mind', so to speak, in spite of misconceptions in its 'conscious mind', that is, by the operations handled by the highest control element in the robot.
Later in the book, so-called 'Machines', with a capital M, are introduced that are a cut above ordinary robots. They are ultra-intelligent and are more or less in charge of groups of countries. A subtle difference now occurs in the interpretation of the first law which becomes (p. 216) "No machine [with a capital M] may harm humanity; or, through inaction, allow humanity to come to harm". And again "...the Machine cannot harm a human being more than minimally, and that only to save a greater number".
Unfortunately it is easy to think of circumstances where it is necessary to harm a person very much: for example, in the allocation of too small a number of dialysis machines to people with kidney disease.
Asimov's book has the important message that intelligent machines, whether they have an ordinary status or are ultra-intelligent presidents, should be designed to behave as if they were ethical people. How this is to be done remains largely unsolved except that the flavour is utilitarian.
The problem splits into two parts. The first is to define what is meant by ethical principals, and the second is to construct machines that obey these principals. |
