Understanding AI Ethics: Part One
Based on the YouTube video ‘AI Ethics’, created by Dr Giles Cuthbert, Managing Director at the Chartered Banker Institute.
Before examining the relationship between ethics and artificial intelligence (AI), it’s important to consider ethics itself and the constructs we have which support our theories of ethics. When thinking about ethics, we’re essentially talking about how to make a judgement between the ‘right’ and the ‘wrong’ thing to do. Here, we must understand that all decisions are ethical. Often, we think of only challenging questions being ethical, but for every decision, a person will always subconsciously think, “am I doing the right thing or am I doing the wrong thing?”. This becomes habitual, and therefore we simply do things which we know do not cause harm to others and are therefore good in some way.
We only truly stop and think consciously about a decision when we think we might be doing the wrong thing: that is an ethical dilemma. However, that doesn’t mean that all things we naturally do aren’t also ethical questions. It’s simply that, for some of these dilemmas, we have to stop and think for longer about our decision, and that is essentially what is at the heart of ethics – what do we do when we’re in that ethical dilemma situation?
The Importance of Capacity, Agency, and Responsibility in Ethical Decision-making
One of the key features of an ethical situation is if the person actually has the ability to take an ethical decision. For example, a 3-year-old child would not be expected to make ethical decisions about complex matters - or even basic matters – as they do not have the ability to do so.
This pinpoints the concept of capacity in ethical decision-making. Capacity is a precursor to responsibility - that is, in order to take responsibility for something, we have to have capacity to make the decision in the first place. There are different forms of incapacity, such as a mental illness, age, or a number of other factors.
Once it has been established that a person has capacity in an ethical situation, they then have the ability to take a decision that is seen as valid: this is what we describe as agency. In the context of a professional, agency refers to their capacity to take a decision in a certain situation for which they then have responsibility. So, once we know there is agency, the next step is identifying responsibility.
When identifying responsibility, there are two key features which must be considered: firstly, whether the person has sufficient knowledge of the situation, and secondly, whether a person has control of the situation (or certain aspects of the situation). This is important, as a person can be involved into a situation but not have control over the outcome, and therefore cannot be considered to be responsible.
It’s that first factor of sufficient knowledge where professionalism comes into play, because only professionals can take responsibility in certain situations where professional knowledge is required. For example, an amateur may be involved in an ethical situation, but they could not be seen as responsible as they don’t have professional authority or knowledge needed for that situation. It is this factor of responsibility that allows for praise and blame in certain ethical situations.
It is these features of agency, professionalism and responsibility which allow us to understand the distinction between good and bad; this can also be referred to as a value-based (or axiological) distinction. And our role as a professional body is ensuring that we have individuals that are confident in their role as the responsible human to call on their professionalism, expertise and judgement to look at the distinction between good and bad outcomes i.e. is it better or worse that I do this than an alternative action?
The concept of ethical decision-making in relation to AI ethics, centres on how digital systems and AI take automated decisions throughout the course of their daily operations. The key distinction is therefore on the unique ability that humans have to intuitively know all these things they’re doing are good on a normal, day to day basis. When humans are struck by an ethical dilemma, their identification of that ethical dilemma is triggered by their ethical sensitivity - the human capacity to notice that there is a problem or challenge that requires particular ethical attention or analysis. In other words, it is our ethical sensitivity that causes us to stop and think about some things more than others.
Although this may sound simple on the surface, there is a significant history of ethical dilemmas and scandals caused by organisations or individuals who did not stop long enough to reflect on whether there was an ethical problem with something they were attempting to do. It is this failing of ethical sensitivity which lies at the heart of many ethical problems.
To learn more about AI Ethics, look out for Part Two of our AI Ethics blogs on the Institute blog page or watch our YouTube video here.