The Ladder of Artificial Agency

"Civilization advances by extending the number of important operations which we can perform without thinking about them" Alfred Whitehead

As I write these words – or rather, as an AI assistant collaborates with a human author to write these words – we're engaged in exactly the kind of agency-sharing that this essay attempts to think through and classify. For 150,000 years, humans have prided ourselves on being the exclusive wielders of rational, deliberate agency in our world - or at least we liked to think so. Other forms of life have always exercised their own kinds of agency, from bacteria navigating chemical gradients to birds crafting elaborate nests and beyond. But we told ourselves our particular flavor of consciousness and decision-making was special. Now we're building artificial minds that can act with increasing autonomy, and the most interesting question isn't how intelligent they are, but rather what kind of agents we want them to be[1].

To understand the different modes of artificial agency – the different ways that intelligence manifests as action in our world – we can imagine them arranged on a ladder. Not an evolutionary ladder (these modes coexist and always will), but rather a ladder of agency distribution: how much of the total agency in any given action stays with the human, and how much transfers to the artificial system[2].

Leave a Comment