Engaging with ELIZA
Earlier I spoke about ELIZA and argued that ELIZA does not in fact have a mind.
The Turing Test looks at ELIZA’s objective behaviour - her responses to questions posed - rather than the subjective thoughts underlying these responses. I suggested that the Turing Test asks the wrong question. Instead of asking whether ELIZA has a mind or not, we should ask whether ELIZA has subjectivity.
Test for yourself whether you think ELIZA has subjectivity. From experience interacting with ELIZA or other AI, how do you go about testing for a subjective mind? Does this suggest any limitations of Turing’s behaviourist approach to test for a subjective mind?
ELIZA now seems a little antiquated given all the AI being used. This has one advantage for us - it is much easier for anyone to see what the mechanism is behind this ‘engagement’. ELIZA was originally created by Joseph Weizenbaum in 1966. He did not claim ELIZA has a mind or that it passes the Turing Test, nevertheless, it has become a widely used example to illustrate these issues.
© University of Cape Town CC-BY-NC