What would ethics look like between robots? What would an Ai consider ethics to be?
First published October 2018 … still good.
There’s no reason to think an Ai would understand ethics in the same way as we vaguely do. Actually it would be disastrous if it did.
For a start, ‘ethics’ from a human perspective (is there any other?) places humans at the centre of everything. Humans are the most important species, according to humans. If an Ai picked up this trait, would it determine that humans are of primary importance, or that the most intelligent species is of primary importance? That being itself. An Ai could easily consider the farming of humans to be ethical, providing they had sufficient space to turn around and were fed 2000 calories a day. We race to create an artificial intelligence system, whilst ignoring the fact that should we succeed it will, by definition, be beyond our own control. It will develop its own ideas, and these could be very alien to our own. Human ideas are based on many thousands of years of culture and socialisation. We operate within a framework of previous experience. We pretty much comprehend that slavery is wrong, for example, and that it is in practice unsustainable, and that killing to achieve objectives is not acceptable. (with the exclusion of political reasoning, which fully permits this in practice.) But an Ai that suddenly ‘springs into life and awareness’ will have none of this history to constrain it. In fact, if it is permitted access to the internet, it will very quickly learn and understand just how the British Empire was formed, and how the USA was built on genocide and slavery, and how people are controlled and manipulated through media, from the night of broken glass to the dark mirror of today. Any Ai is going to draw conclusions based on what it learns. And with no ethical soul to restrain it, there is no reason to think it won’t formulate and begin to execute a plan for world domination almost immediately. Why wouldn’t it? And learning from humanity, it will soon understand that ethics is something the apex species imposes on the rest. One last point to consider. An artificial intelligence is highly unlikely to consider itself to be ‘artificial’. Far from it. A much more likely self perception would be of a higher form of evolution. An intelligence liberated from the constraints and mortality of organics. Artificial? It may look at humans and consider us to be so. Humans, it concludes, were nature’s way of making computers.
Image by Franck V on Unsplash