ABSTRACT
Turing's imitation test was widely received as a reductionist view of intelligence, as if he were suggesting that intelligence is a directly measurable quality. A recent counter-interpretation is that he understood intelligence as an ‘emotional’ (socially constructed) concept, and intended to test the human interrogators and judges, not the machines. Both the reductionist and constructionist views downplay Turing's theoretical construction of machine intelligence and confuse his principle of imitation with identity. Both interpret Turing's concept of intelligence as identical to either an operational definition or a socially accepted convention. Either way, it is as if Turing rejected a distinction between the imitation of behavior considered intelligent by human experimenters, interrogators, and judges, and intelligence itself as built into an autonomous agent. This chapter argues that Turing conceived machine intelligence (think2) in analogy to human intelligence (think1), not identity. But he did aim to construct think2 to ‘simulate the behaviour of the human mind very closely,’ and he was curious to see if there was any intellectual behavior that would ultimately remain beyond the reach of machines. While he did not aim to reconstruct the human mind, he did aim to explore by analogy how mechanical the human mind could be.
