The second is a combination of language and motor skills In 5 Harnad proposes

The second is a combination of language and motor

This preview shows page 6 - 8 out of 10 pages.

The second is a combination of language and motor skills. In [5], Harnad proposes the “Total Turing Test” (TTT), which requires a machine to possess both of these skill sets. As pointed out by Hauser in [6], Harnad actually states in his paper that linguistic capability tends to be evidence of motor skills. Thus, the TTT is redundant: if a machine could pass the TT, he could pass the TTT, assuming Harnad’s conjecture is correct. In fact, Harnad goes so far to claim that body and mind are likely inseparable, so a test for only motor skills would also be redundant. 6
Image of page 6
6.893 Katrina LaCurts Just to address a misconception of the TT (that Harnad seems perhaps to have), nowhere in the test did Turing say that linguistic skills were the only skills the machine would have. If body and mind are in fact inseparable, then yes, we will not be able to create intelligent machines that have linguistic skills and not motor skills. But that is an implementation issue; it does not speak to a defect in the test. Finally, Schweizer [15] goes one step further than Harnad, claiming that machines should have linguistic abilities, motor skills, and the ability, as a species, to create. He proposes the TTTT (“Truly Total Turing Test”), which requires an entire specifies of robots that evolve and create, and is in some sense attributing intelligence to a species, not to an individual. While that may be a fine extension of the TT, it seems di ffi cult to generate a species of robots that could pass the TTTT without starting with a single robot that could pass the TT. 6.5 Summary of These Criticisms To summarize this section, most criticisms of language as it is used in the TT claim that language does not capture all manners of intelligence. However, language is how we test for intelligence in humans, and we arguably do a less thorough job of testing when we converse with a new human than we would with a new machine. Additionally, it is not clear that language cannot capture all types of intelligence, as evidenced by my argument against Michie [9] and Hauser’s [6] criticism of Harnad [5]. If, one day, we were to find a definite type of intelligence that could not be captured by language, then yes, we may want to develop a new test for intelligence. But I would argue that we must work on building intelligent machines before we can discover if this distinct type of intelligence exists (in part because it will be di ffi cult to isolate any portion of intelligence in humans who already have linguistic abilities), and also that any new test for intelligence would likely be a stepping stone on the way to passing the TT. After all, though it would be a great achievement to pass this new intelligence test, our machine would likely be missing some key intelligences that are captured by the TT.
Image of page 7
Image of page 8

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture