Kantian Agency and the AI Question

A person, according to Kant, is one whose actions can be “imputed” to him and who is “subject to no other laws than those he gives to himself,” whereas a thing is an “object of free choice” to which nothing can be imputed (MM, 6: 223-4; Kant 1996, 378). If Artificial Intelligence (AI) has evolved today to the extent that it can simulate human intelligence in decision-making, learning and even understanding language, it invites an assessment whether it is capable of giving laws to itself and if it has the ability to act in accordance with such laws. Kantian agents are capable of making “free” choices, the character of which can be understood by “a theoretical presentation of freedom as a noumenon” – that is, “freedom as the ability of the human being as an intelligence” (MM, 6: 6:226; Kant 1996, 380). AI must show the same “ability” if it claims agency for itself, else it will remain a “thing.” In this paper, I will explore the relationship between human intelligence and artificial intelligence further. I will examine if the relationship between the human being and AI can be understood as one of two persons, or of a person’s right to a thing, or as a right to a person akin to a thing.