At the beginning of the month, I set the stage for my next several posts by engaging with a lecture by Notre Dame professor Meghan Sullivan on the topic “Ethics, AI, and Human Flourishing.” During that lecture she introduces a framework for thinking about AI in the context of Christian values and virtue ethics that goes by the acronym “DELTA,” as described in the following short video:
Last Sunday I focused on the fourth letter in the DELTA acronym: “T” is for Transcendence. Today’s letter: “A” is for Agency.
Agency places us squarely in the playground of freedom and responsibility. Not “freedom” in a political sense, nor “freedom” in the sense of being able to do whatever I want whenver I want without anyone stopping me, but something more basic and existential. When philosophers talk about human freedom, they are invariably referring to free choice. A quick definition: My choice to do “A” is free if among various alternative possible actions I choose to do “A”–but I could have done otherwise. A free action is the product of a free choice. I deliberated between the possibilities, weighed them in various sorts of ways, then chose–but no one or nothing forced me to.
Whether human beings actually have this power of free choice or whether we simply think we do has been a matter of debate amonst philosophers, scientists, theologians, psychologists, and anyone else who wanted to join in for centuries. I’ve written about the issue several times over the years in this blog–here’s a foray from three years ago:
The issue of human free choice intersects directly with AI when we consider the other side of the free choice coin, the side we often choose to conveniently ignore–responsibility. Everyone likes the idea of choosing freely, but far less popular is the idea that if I choose freely, then the results of the action, for better or for worse, are my responsibility. If the action was truly up to me, then I cannot pretend or deny that I was not responsible for it. Our whole moral framework of reward and punishment, as well as our legal system of innocence and guilt, are built on this foundation.
Issues of responsibility frequently arose during the recently completed semester in my “We Can . . . But Should We?” colloquium. For example, we spent several classes investigating how AI is being integrated rapidly into defense and security operations. The use of “killer drones,” for instance, has become central in the Ukraine-Russia conflict. Ukraine, vastly outnumbered in both manpower and equipment, has been able to not only defend itself but occasionally go on the offensive against a larger foe through drone warfare, often with drones cobbled together from spare parts taken from other sorts of technology. We know that drone warfare has made it possible for devastating strikes to be made against various enemies as buttons are pushed by decision makers thousands of miles away.
But what happens when AI systems and technology develop to the point where AI itself makes the choice of which targets to strike and with what sorts of deadly force? My students were convinced that AI does not and cannot choose freely as we believe human beings do, but it can generate actions for which we will want to be able to assign responsibility. In her talk, Meghan Sullivan asks the question my students were asking throughout the semester: Can AI be morally responsible for decisions that do not directly involve human beings?
If free choice and responsibility are inextricably linked, and if AI doesn’t “choose” in the sense that we believe human beings do, then the answer to the above question seems to be “no.” And that’s a big problem. Humans have a poor track record for taking individual and/or collective responsibility for things that go badly–AI undoubtedly is opening the doors wide to a future in which this will be much easier to do, while at the same time being very happy to take responsibility for things that turn out well.
Issues of responsibility in the world of AI are not, of course, limited to defense and security issues–they are pervasive in all parts of our daily lives. Perhaps a good place to start is to simply aske whether there are some decisions that only a human conscience should make. What are they? Can we build identifaible and agreed upon guardrails around AI that protect and preserve free choice and responsibility? These are conversations that cannot be avoided.
I close with this anecdote from a different class this semester. My teaching colleague was on point that morning, facilitating a discussion closely related to the above issues. Part of the conversation raised the question of whether AI can “think.” On the spur of the moment, I asked ChatGPT the following question: “Does God exist?” Within seconds, it answered
People have argued both sides for millennia, and there isn’t a definitive proof that convinces everyone. It comes down to which kinds of reasons you find most compelling,
Followed by multiple bullet points summarizing different arguments for and against that have been offered ovr the centuries, more or less the same quality as the proverbial student paper that starts with “Since the beginning of time . . .” No kidding–I’ve taught Philosophy of Religion before. My follow-up question was what particularly interested me: Do YOU think God exists?
I don’t have personal beliefs or a inner life, so I don’t “think” or believe in the way people do. I can outline perspectives and evidence, help you evaluate arguments, or simulate how different traditions or philosophers would respond. If you tell me what standard of evidence you care about—scientific, philosophical, or experiential—I’ll assess the case through that lens.
That was encouraging–no delusions of being a person (yet). Most interesting is AI’s admission that it doesn’t have an inner life. After five consecutive posts on the DELTA framework, that might be one of the most important connecting threads. The inner life is what distinguishes humans from everything else, including what we create. If there is anywhere that we reflect what is greater than us, that would be the place. Let’s not forget it.
Ad-Free & Subscriber Log-In