Abstract

In this paper I identify and discuss a number of features that I argue are necessary for the realization of AGI. As a preliminary step, common definitions of AGI are presented in respect to their understanding of mind, intelligence, and consciousness. I show that, despite the amazing performance of artificial systems, at present they are still far from exhibiting AGI, and I identify some of their central short-comings. Secondly, inspired by research within the philosophy of mind, embodiment and situatedness, I suggest a number of features that I deem necessary for a mind. I then investigate the possible objection against the relevance of these features namely that they are overly anthropocentric or biocentric. I further discuss aspects of these features in relation to their transfer to artificial systems with the goal of creating an artificial mind. I finally conclude that self-reflexivity and the re-creation of the world as an inner world should be strongly focused upon if one wishes to create an artificial mind or artificial consciousness. However, I also issue a warning about some well-known risks when creating AGI.