Reimagining Artificial Intelligence through the critical lens of Harry frankfurt

Harry Frankfurt’s philosophy of “free will” presents a significant departure from classical compatibilist, in-compatibilist, and libertarian accounts of free will by shifting the focus from alternative possibilities to the structure of human agency. Frankfurt argues that what matters for a person to have free will1 is not the ability to do otherwise, but the capacity of reflective endorsement of one’s desires, a hierarchical model of the will. He distinguishes first-order desires from second order desires, suggesting that, in order to possess free will an agent must have alignment between these levels of desire and its volition. This reflective structure of Frankfurt philosophy is central to the concept of personhood and free will. Frankfurt defines personhood in terms of capacity of an agent to have second-order volitions, especially the ability of reflective self-endorse of desires. Frankfurt’s conception of personhood is framed in psychological and volitional terms, not strictly biological or species-based. So, in principle any being biological or artificial that possesses second-order volitions can qualify to be a person in Frankfurt’s sense. In the philosophy of mind, the prominent theory of “functionalism” holds that mental states are defined by their functional roles, how they interact with and what role they play, not what they constitute of. This “functional” and “hierarchical” conception, theoretically, allows for non-human persons, including AI, to possess intelligence, cognition and free will, if they have the right kind of structure, what matters for intelligence or personhood is the pattern of operations, not the material from which the being is made. This paper explores the possibility of attributes such as mental states, cognition, free will and intelligence, which are exclusive to Human beings, in Artificial Intelligence machines and critically evaluates the functionalist view of mind with reference to Frankfurt’s conception of personhood and free will. This paper aims to initiate an inquiry into the extent to which attributing features of human cognition to artificial intelligence is justified.