“Ultimately, nothing or almost nothing about what a person does seems to be under his control.” 

                                                                               - Thomas Nagel, Moral Luck

We look, critically, at how our technologies work, and yet we make assumptions about how we work. What motivates our choices? Are we in control of our actions – and if so, all of them, or only some of them? As our interactions with and dependence on new technologies, including AI, become both increasingly common and invisible, what, if any, agency are we giving up? If we better understand our agency, how does this connect to our responsibility for the technological world we are creating, and the natural world we are destroying? What responsibilities should we have for our own behaviors, and where does accountability reside in automated systems?

We use the term “agency” to refer to humans, to current and future AI systems, as part of a framework for responsibility and accountability. But what do we mean by agency? Agency is defined differently across disciplines -- from computer science to philosophy to sociology to law. Recent developments in neuroscience and AI both call into question the accepted notion of volitional agency as the willed proximate cause of a thought or an action. How might exploring frameworks of agency affect our approaches to ethical standards in the development of AI? A potential blind spot in our analysis of the development of AI lies in the assumptions we make about our own agency, freedom of will, and moral capabilities.

Are we are actually more accurate when describing the behavior of machines -- mechanistic, physical, governed by the laws of nature and programming -- than we are when we describe ourselves? Things get fuzzy as the mysteries of consciousness and subjectivity arise. What is true – and what, if not true, is useful to believe?

We believe that we, as humans, have at least some agency. We acknowledge that our degrees of agency differ across individuals and circumstances, increasing or decreasing based on certain constraints, and governed by physical laws -- at least those outside of our brains. Most people don’t believe that they could defy physical laws: the laws of gravity, survival without food, etc. We accept these physical constraints, those that appear to affect all beings and appear to be external to us, or at least external to our physical bodies. Yes, this agency is highly variable: a healthy adult has more agency, people tend to agree, than a baby, or someone who is very old or unwell.

We tend to agree that we do not have the agency to fly, or to travel in time, or countless other fantastical things (barring of course certain mental illnesses, or other illnesses which impinge on mental capacities, which have their own unique relationship to agency and thus also to responsibility). And yet most people now, and throughout history -- across cultures, ages, and every other demographic factor -- have a distinct sense of being in control of (at least some of) our behaviors and actions. Even though it is difficult to explain, there is a distinct and overwhelming sense that I am choosing to write these words, that I will choose what to have for dinner, that I could choose to clap my hands, or nod my head, or close my eyes. This sense, as inexplicable, biologically and physically, as it may be (as a being comprised of physical matter that came into existence in a way that I certainly did not will), from where did it arise? Is the sense of agency I possess merely a myth? Perhaps a useful, or even inescapable myth? If so, is considering such questions useful or productive?

For me, reflecting on such questions is enriching: it enriches my daily life and my experiences. Paying attention to this deep and abiding mystery, somewhat ironically, feels empowering – as if I am curiously contemplating whether the backdrop is a facade, whether this sense of agency is indeed an illusion. I acknowledge the possible privilege of this perspective. Perhaps, if I do indeed have some sort of inexplicable agency, contemplating it is enjoyable because I have (if I have it at all) a relatively high degree of it. But perhaps not.

Such topics have fascinated philosophers, theologists, and most humans for so long as we have records of such contemplation. Debates on free will or the existence of agency -- nevertheless have barely made their way into discussions of the new sophisticated technologies we are creating -- particularly AI, in terms of how it already is acting in the world, as well as how it could impact the future. We talk about autonomy and responsibility, but can we use this moment to also reflect back on our assumptions about ourselves?