Artificially Intelligent Reasoning and Human Responsibility

July 9th, 2019

 

I have recently been involved in exploring the application of what is widely called artificial intelligence or, more appropriately for the most part, machine learning (that is, computers programmed to learn, recognise and respond to patterns) to some of the challenges that confront us in modern nations. In describing the emergence of contemporary cybernetics in the 1940s (some of the techniques from which are now embedded in these machine learning capabilities), Wiener wrote of ‘the initiation of a new science which, …, embraces technical developments with great possibilities for good and for evil’ and we are now seeing those possibilities made real.

Almost everybody that uses the internet will have encountered such applications. It could be Google promoting events near your location (yes, they generally know where you are), Amazon telling you that ‘people who looked at A also looked at B’ (people who bought a hammer often bought nails), Facebook presenting advertisements based on your profile and history (assuming that you should be interested in the same things as everybody you are alike and/or connected to) or a myriad of other ‘aps’ busily learning away. In fairness, the patterns are sufficiently accurate to make the business models work, if not then they wouldn’t bother (and yes, we generally are that predictable)!

These are the realities of lives lived on the internet and may be considered beneficial and benign (Oh, marvellous, I have always wanted a…) or damaging and malign (I feel like I am under surveillance, why are they always trying to sell me stuff, what makes them think I am interested in…). Of course, the challenge arises that it is not just our shopping habits that machines can learn about, they can learn the patterns of our lives (to turn the heating on or off), the times when we are on (or off) the internet and the locations from which we gain access – they can learn where we are, who we communicate with, where we bank and a whole raft of other matters. These patterns, and we all have them, generate the opportunity for ‘bad people’ to take advantage of us.

Of course, the computers themselves and the algorithms which they run are neither benign nor malign, they do not act to benefit or harm us, they simply follow the instructions of their human programmer. Whilst some may be capable of representing emotion the ability to do so is part of the algorithm, built in as a conditional response by the programmer, not (yet?) developed as part of a machine psychology! Perhaps we impute benign or malign intent as a reflection on our emotional response to the experience of using the application or machine; then how we feel is NOT about the machine but about OUR behaviour in relation to the machine?

‘I have spent too much on xxx again this month and now I am in debt, I wish the machine would stop selling me things’ is to displace the blame for our behaviour and avoid responsibility. Similarly to be grateful, ‘if it had not been for Amazon I would never have this perfect product,’ is to displace the credit for our success and thereby not celebrate OUR capability in searching! We only ‘ask Google’ when we think that is the most efficient way of finding the right answer.

Perhaps WE need to take responsibility for the good and the bad, for the benign and the malign, to think differently about the underlying causes of our various experiences. We must not only develop healthier relationships with the baronies and fiefdoms of the internet but engage in behaviours that avoid damaging activity. We need to teach ourselves and all those for whom we have responsibility or a duty of care to be philosophers, to reason about and take responsibility for the use of technology, to make and reflect upon choices and become active, informed, buyers of beneficial goods and services not passive consumers of the amoral pap arising from mechanistic algorithmic machinations.

Remember, at least for now, everything which is learned by a machine is a result of an instruction built into it by a human being. Next time you explore or buy, don’t just think about the information, product or service but about the extent to which the motivations of the person behind the algorithm and YOUR motivations are in alignment!

The internet is amoral, it is benign or malign as a function of its users.