• tuesday, 19 september 2023—12:15

    Aïda Elamrani - To what extent are machines conscious?

    Aïda Elamrani, Institut Jean Nicod, ENS-PSL

    As AI improves rapidly, the question of whether Machine Consciousness (MC) and the extent to which this is possible becomes increasingly crucial. The debates surrounding the hard problem expose a stark disagreement over the possibility of a strictly mechanical implementation of consciousness among experts of this interdisciplinary field. I first give an overview of the challenges facing MC by explaining the arguments opposing phenomenalism to physicalism. Evaluating the possibility of MC requires an in-depth analysis of information, computation, and their place in nature.
    Information and computation have shaped several metaphysical theories. Underlying the “IT from BIT” view is the idea that reality constitutes a natural pool of information. Meanwhile, the scientific study of consciousness describes our minds as natural information processing devices. Together, these two bodies of work avail a compromise to close to gap: phenomenal experience as a virtual reality, physically implemented through layers of computational mechanisms. I show this compromise motivates further research questions between two interpretations of information: idealism and pragmatism.
    In practice, this analysis sets two distinct goals for MC: strong (phenomenal) MC and weak (functional) MC. Which one of them is achievable, and to what extent, depends on the interpretation of information we endorse.

    Teams link: https://bit.ly/3r3uYOx

    external seminar