What Counts as Agency in Artificial Systems? Bats, Technological Mediation, Digital Dinosaurs, and the Space of Reasons

This article asks what should count as agency in artificial systems. Contemporary discussions of artificial intelligence often move too quickly from goal-directed performance to stronger attributions of autonomy, intelligence, or human-like agency. I argue that a crucial distinction must be drawn between pursuing goals within a given framework, responding to considerations as reasons, and participating in the constitution and revision of the normative environments in which reasons and goals acquire significance. The argument begins with Thomas Nagel’s distinction between functional equivalence and subjective experience and develops through Don Ihde’s account of technological mediation. The historical cases of William Herschel’s detection of infrared radiation and Paul Langevin’s development of ultrasonic detection show how technologies can transform human perceptual relations without eliminating phenomenological difference. The analysis then turns to The Isle, a multiplayer survival game in which human players act through computationally constrained non-human bodies. The game makes visible a distinction between computationally specified conditions of action and reasons generated through social interaction. Finally, drawing on recent debates on artificial agency, including Luciano Floridi, Leonard Dung, Federico Faroldi, and Mandy Zafar, I distinguish goal-directed agency, reasons-responsive agency, and normatively transformative agency. Artificial agency may be genuine without reproducing human consciousness or intelligence. A further and distinct philosophical question arises, however, if artificial agents become capable not merely of pursuing objectives or responding to reasons, but of participating in transformations of the normative space within which objectives become meaningful.

کلیدواژه ها:

Artificial agency ، Reasons ، Normativity ، Artificial intelligence ، Technological Mediation ، postphenomenology