Susan Blackmore on machine ethics

Susan Blackmore weighed in recently on the topic of machine ethics - in article titled: It’s too late to give machines ethics – they’re already beyond our control.

While I generally think that students of cultural evolution ought to be well placed to contribute to prediction of the consequences of our actions in this area, I found several things to object to in Sue's analysis. On problem is the article's title. It strikes me as being defeatest. We ought to at least try.

Later in the article, Susan writes:

Replicators are selfish by nature. They get copied whenever and however they can, regardless of the consequences for us, for other species or for our planet. You cannot give human values to a massive system of evolving information based on machinery that is being expanded and improved every day. They do not care because they cannot care.

This seems like confused reasoning to me. We have examples of companies, governments and other organizations which have codified various human values. These are often in the form of 'laws' or 'rules'. If the argument is that memes and genes 'cannot care' because they are selfish replicators, then we have many examples of complex meme or gene products which do care - or behave as though they care.

Blackmore looks as though she is arguing from selfish memes to selfish organisms here. If so, that is a mistake parallel to the mistake that Richard Dawkins made in The Selfish Gene. Dawkins (1976) wrote:

I shall argue that a predominant quality to be expected in a successful gene is ruthless selfishness. This gene selfishness will usually give rise to selfishness in individual behavior. However, as we shall see, there are special circumstances in which a gene can achieve its own selfish goals best by fostering a limited form of altruism at the level of individual animals.

...and then...

Let us try to teach generosity and altruism, because we are born selfish.

He subsequently had to back-pedal, writing the following belated retraction (2006):

I do with hindsight notice lapses of my own on the very same subject. These are to be found especially in Chapter 1, epitomised by the sentence ‘Let us try to teach generosity and altruism because we are born selfish’. There is nothing wrong with teaching generosity and altruism, but ‘born selfish’ is misleading. In partial explanation, it was not until 1978 that I began to think clearly about the distinction between ‘vehicles’ (usually organisms) and the ‘replicators’ that ride inside them (in practice genes: the whole matter is explained in Chapter 13, which was added in the Second Edition). Please mentally delete that rogue sentence and others like it.

Selfish memes could result in selfish companies, governments and organizations - but it ain't necessarily so. Saying that such complex entities 'cannot care' seems like an unwarranted generalization to me. Maybe memes "cannot care" - but so what? It is memeplex products that we are mostly interested in when discussing machine intelligence. There's no good reason why they can't care.

Anyway, this topic is an important reason to study memetics. We need the best science has to offer to help us predict the consequences of our actions. The existing man-machine symbiosis probably won't last forever - there will probably be a merger or one side will assimilate the other. It looks as though we have enough power to be able to influence the outcome in a variety of ways - to the extent that technological determinism leaves some aspects of the outcome open. There's a lot at stake - and we should try our best to figure out what we should do.