Artificial intelligence and the Kantian antinomies

Abstract

In this chapter we will analyze the epistemological question "Can we know if an AI can have a mind?" as if it were a fifth antinomy like the ones of Kant's "Critique of Pure Reason" (Kant, 1998). This thesis is reflected in Turing's position in 1950, in accordance to the field of Strong AI, by passing the Turing Test. We have another answer: Alan Turing, in 1939 still agreed with (Gödel, 1931) results on the incompleteness theorems He had arrived at the same results with the Turing machine (Turing, 1936). Turing extrapolated the formalization in mathematics and logics into the field of computation, and wondered about "intuitions" in machines. He presented a "System of logic ordinals" (his Doctoral Thesis at Princeton, directed by A. Church) with negative results, which showed that he could minimize but not remove all external intuitions (external semantics) from a system, even though it got more complex. In the time→ +∞ we would remove external intuitions and we could know if it had a mind (therefore we can never know- antithesis). Besides, there is another thesis in this case: the biologistic point of view of Searle (1980). It is a thesis because it states that biological beings are, by being biological, especially high species of animals (like macaques and corvids) (Hauser, 1997; Santos et al., 2006) and human beings, endowed with some sort of consciousness.