Turing offers what’s now known as the Turing test as, in effect, a definition of machine intelligence. But a definition in itself is empirically meaningless. Is the TT as Turing describes it an adequate test for human-like intelligence? This is a question where intelligence is not defined as what a computer does, but where “intelligence” has a meaning independent of what computers do, and therefore what computers do can be subject to empirical testing to determine if they are – or are not – intelligent.
On reading Turing, it quickly becomes clear that the Turing test is in fact a test of human-like intelligence. The responses of the computer to typed questions are compared to the responses of a human. If the interrogator can’t reliably tell the machine response from a human response, then the machine is deemed intelligent. That means human-like intelligence. Hence to regard the TT as a definition of intelligence seems irrelevant to the question: Could a computer think?