No Algorithm for Self-Awareness On the Human Exclusivity of Awareness of Awareness

Summary

Self-awareness in AI cannot be achieved because it requires subjective experience and a unity of consciousness that AI systems lack. The paper examines Brentano's concept of inner perception, where awareness of an object simultaneously includes awareness of the mental state itself, and Kriegel's self-representationalism, which emphasizes a first-personal character or 'for-me-ness' in conscious states. AI's predictive outputs, being devoid of subjective experience and consciousness, cannot replicate this awareness of awareness. Therefore, no algorithm can produce genuine self-awareness.

Study at a glance

Abstract

This paper addresses the question of whether AI can have self-awareness in the Brentanian framework of inner perception and Kriegel’s view on self-representationalism. First, I will discuss the notion of inner perception in Brentano, showing that, when one is conscious of the object of consciousness, at the same time, she is conscious of that mental state, and there is a subjective unity. Kriegel, with his same-order self-representationalism, strengthens Brentano’s notion of self-awareness. Both talk about a unity of consciousness which has for-me-ness. Then, I will show that the predictive output of any AI system cannot be compared with such awareness of awareness because it is devoid of the subjective experience, and it lacks consciousness altogether. Hence, there is no algorithm for self-awareness.