傷つかないAI、傷つく人間

抄録

The discomfort people feel toward judgment and governance by AI is examined in this paper. Firstly, a review is conducted concerning existing discussions in law and philosophy, which highlight concerns about the dangers of reducing individuals to bundles of attributes and processing them probabilistically, as well as the unease stemming from the way AI recommendations and decisions contribute to the construction of the self. However, such phenomena can also occur when the decision-maker is human. An additional issue that arises in the case of AI is its lack of vulnerability. Because AI cannot be harmed, it becomes difficult to conceptualize the attribution of responsibility. This phenomenon may underlie the discomfort people experience toward AI.