AI leadership needs practical wisdom
Photo credit: Cliff Johnson via Unsplash
AI ethics must be more than a set of policies, checked off through a compliance process or reviewed in a governance meeting. Ethics must include reflection, opinions and never-ending arguments about how we should live, and AI ethics must include patterns of thought and behaviour which we deliberately, and deliberatively, seek and acquire.
Making sense of AI is difficult, especially if we take the task seriously: that is, not just attempting to understand the engineering and mathematics of AI, or even its application and economics, but also its impact on our minds, knowledge, roles, relationships and all the other things that make up our lives. I believe that a philosophical tradition over two thousand years old can help, as long as we approach it thoughtfully.
Of the various leading schools of ethical thought, I favour the category labelled virtue ethics. Unlike other ethical theories popular in Silicon Valley, such as consequentialism (the theory that we should seek the greatest good for the greatest number, which appeals to optimising engineers), deontology (the theory that we can there are universal, objective moral rules, which appeals to programmers seeking categorical distinctions) and Stoicism (a doctrine of discipline and self-reliance, which appeals to entrepreneurs who like to think about the Roman Empire), virtue ethics is messy, unfinished, contextual and, I think, manages to capture some of the ethical complexity of actual life.
The origins of virtue ethics are often credited to the Greek philosopher Aristotle, who lived in the 4th century BCE, although other forms of virtue ethics and other origins are available (through, for example, Confucius or the traditions of Buddhism: for a fantastic exploration of these ideas and their relevance to technology, see Shannon Vallor’s book, Technology and the Virtues).
It’s always dangerous to follow all of the ideas of any particular thinker unquestioningly, and there are a lot of things to question about Aristotle, including the attributes that come from being part of a slave-owning, male-dominated society, as well as some of the details of his ideas about virtue. However, I think that there are some ideas which are particularly valuable, such as that: ethics is primarily about character, the sort of person that we should aspire to be; character is expressed through action; a good character manifests virtues, admirable traits such as courage, generosity and honesty; the practice of virtue reinforces itself (we become more honest the more we practice honesty and so on); virtues have corresponding vices, such as cowardice, miserliness and dishonesty; just like virtue, practicing vice reinforces itself; and practicing virtue is essential to human flourishing.
Exploring all the features of Aristotle’s conception of virtue, as well as its flaws, would make this a very long article indeed, and has already been attempted by better people than me. However, I think that there is one aspect of Aristotle’s virtue ethics which is particularly helpful in making sense of AI: the idea that, as well as moral virtues, honesty, courage and so on, there are intellectual virtues, representing excellence in reasoning.
Aristotle identifies five intellectual virtues; episteme (knowledge), techne (art or technical skill), nous (intelligence or intuition), wisdom (the combination of deduction and intuition to discern universal truths) and, I think the most relevant to our use of a tool such as AI, phronesis, or practical wisdom. The idea of phronesis may be intensely frustrating to those people who come to philosophy looking for categorical rules or simple answers: it essentially says that figuring out the right thing to do depends on the ability to figure it out, and the virtuous person acquires and exercises this ability. The response to the answer seeker will not be: here is the answer, but go ask the phronimos (the possessor of phronesis), or, even better, practice phronesis yourself.
Frustrating though this response might be, it reflects our experience of everyday life. When faced by moral dilemmas we often desire simple rules and certain answers. But, if they are real dilemmas, we know that we cannot commit to a course of action unless we have thought it through. And thinking it through is difficult: it requires us to take a step back from our own interests, to put ourselves in the position of others, to consider wider consequences, to ask what kind of person would act in such a way, and whether we want to be that kind of person. We practice phronesis when we think hard about hard questions. (The frequently missed point of the famous trolley problem is not to find the right answer, but to show that human morality cannot be captured by simple rules, such as save the greatest number or never act to cause harm.
Phronesis is particularly relevant to the work of making sense of AI in two ways.
First, there are many hard questions which require lots of hard thought. We have to make choices about the solutions we are deploying, about privacy, security, intent, accountability, implications for customers and employees, and many other things that can’t be captured in a specification. And we have to make choices about the society and economy we are creating, about control, power, the allocation of resources, the meaning of work, and many other things that can’t be captured in an AI policy statement.
Second, the idea of phronesis and the other intellectual virtues remind us that our minds and the ways that we reason are shaped through practice and habit, and are fundamental to our identities and character. If we are to use AI to assist our reasoning, we must be cautious, conscious and thoughtful about whether it is amplifying our virtues, or encouraging our vices.
Acquiring practical wisdom is hard. Exercising it can be even harder. But preserving and developing our intellectual virtues in the age of AI is worth the work, and essential to human flourishing.
