Abstract

Anthropocentrism and non-anthropocentrism are a pair of basic directions in cognitive science and AI research. These directions correspond respectively to "first person" and "third person". The confrontation between these two concepts not only has guiding significance for the exploration of cognitive science but also has ethical importance. Human-like AI (for instance, Artificial General Intelligence, AGI for short) research hovers between them. I propose a second-person, or limited consensus perspective as a philosophical and ethical research context that is integrated with the first and third person to constitute a "trinity" relationship - what I call the "trinity of the contexts of intelligent artificiality", which could be represented by the relationship between emulation, simulation, and imitation. The consequences of the general loss of the first-person context will be explored, including the uncertainty of the ethical landscape of humankind - such as the fragmentation of the self - and the risk of going against the original intention of AGI research. This cannot be fixed by returning to a first-person perspective. The second-person context is an idea that is situated between anthropocentrism and non-anthropocentrism, and it may be employed to avoid these risks.

Links

• PDF