Abstract:

The evolution of Artificial Intelligence (hereafter “AI”) has opened up an “Age of AI Agents”, in which humans and AI agents mirror each other. AI agents present themselves as “ethical entities” in the society, rationality, value and virtue where they are anchored. Criticizing the computationalist interpretation of AI’s “ethical anchor”, the philosophy of embodiment argues that a computationalist AI agent inevitably falls into the agent traps of “disembodied mind”, “moral machines”, and “meta-act”. According to the general articulation of “intentionality difference”, the comparison between the “intentionality of machines” and the “intentionality of bats” can explain the ethical anchor AI agent. The fourfold tension presented by the intentionality of machines indicates that the AI agent cannot transcend the limit of their initial instrumental form due to the “lack of subjectivity”, and thus there is a hidden “new slavery setup”. As the underlying framework of machine ethics, the co-existence intentionality of humans and machines takes “the self-awareness escape” as a technological approach. However, this actually presupposes the importance of “self-awareness” returning to and reflecting on itself. The worries or anxiety brought by reflexivity is an “ethical surplus barrier” that the intentionality of machines cannot cross. Clarifying the worries caused by “agent traps”, the “new slavery dilemma” and the “self-awareness escape” can help to dispel some ambiguous concepts emerged in the research of AI ethics.

Key words: AI agents, intentionality, ethics, morality