If you imagine an agent as a blackbox, you can first focus on the inputs it can receive and the outputs it can produce. In a second step, you then peek under the hood to understand what happens inside that box, how the inputs are connected to the outputs. Let us begin with the agent’s relations to the outside: What goes in? What comes out?
The agent’s context
Agency can only exist within a given context, and the same applies to agents themselves. These concrete environments or abstract situations can take many different forms, that are introduced here.
With the 'target list' of natural and artificial agents in mind, and focusing on their common characteristics, I can now propose a working definition of agency. At its core, agency is the ability to act. And that action has certain qualities: it serves some purpose, it has a direction towards some utility. Today, I'll offer a short defining phrase and then unpack its meaning piece by piece.
The path(s) to artificial agency
Across the categories of organizations, technologies, and socio-technical systems, we can identify three principle paths to artificial agency. Ultimately, they all originate from human agency; but the degree of human deliberation varies considerably.
Organizations as artificial agents
Unlike the natural agents, which are born, artificial agents are human-made; artificial agents spring from human agency. I will introduce them as three groups that are not perfectly delineated: organizations, technologies, and socio-technical systems. Let's start with organizations, which we use to structure our interactions with others, be that for social, political, or economic purposes.