Abstract

The possibility of artificial intelligence is usually discussed by philosophers as a problem about artificial thought: can an artificial system think? The production of intelligence in an artificial system would, however, give rise to an intelligent artificial being: an artificial thinker. As such, there is another, less explored way of discussing the possibility of artificial intelligence: can there be an artificial thinker? This is the problem of artificial thinkers (Olson 2018). In this essay, I discuss this problem. To do so, first, I present the problem of artificial thinkers and discuss how best to interpret it (section §1). Then, I discuss the three main views on the metaphysical nature of artificial thinkers: the hardware view, according to which artificial thinkers are intelligent artificial systems; the software view, according to which artificial thinkers are types or copies of programs for artificial intelligence; and finally, the CPU perspective, according to which artificial thinkers are intelligent spatial parts of artificial systems (section §2). All of these views have problems regarding their conception of the persistence conditions of artificial thinkers. Even so, I conclude, artificial thinkers are best conceived as functional artificial systems.