Artificial intelligence and the future of otherness: what kind of other can an AI be for a human?

This paper proposes to consider the question "What kind of other can an AI be for a human?", to analyse a set of ethical and societal challenges associated with a hypothetical massive deployment of "AIs as others" in society (social robots, artificial companions, chatbots with relational purposes such as friendship or romantic partnership, etc.). The proposed answer is that six features characterize "AIs as others": serviceable, commodified, surveillant, authoritative, techno-standardized, and interiority-less. The implications of each feature are explored in depth, proposing three possible future scenarios (confusion, degradation, transformation), of which only the transformation scenario would be ethically acceptable. Through this analysis, a central thesis is developed, namely, that one of the main challenges of the transformation that "AIs as others" may catalyse is the adequate understanding of the otherness of AIs in relationship to humans. The conclusion is that for society to develop this adequate understanding, a new kind of "other" must be constructed, tailored to the characteristics of "AIs as others". For it, a new term will be required (it is proposed "analytical other" [AO]), along with a suitable ontological status and a set of admissible roles. This socio-cultural creation of a new kind of other will require a transformation, a more mature and nuanced sense of otherness. The paper concludes by giving some recommendations to bring forth this transformation and proposing some questions and hypotheses that may inspire further research.