Ariella Popple (1996) The Turing Test as a Scientific Experiment. Psycoloquy: 7(15) Turing Test (2)

Volume: 7 (next, prev) Issue: 15 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary
Topic:
Article:
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 7(15): The Turing Test as a Scientific Experiment

THE TURING TEST AS A SCIENTIFIC EXPERIMENT
Commentary on Watt on Turing-Test

Ariella Popple
Psychology Department
Durham University
Durham DH1 3LE England

A.V.Popple@dur.ac.uk

Abstract

Watt (1996) suggests that the conventional Turing Test for artificial intelligence be replaced with an Inverted Test which requires the machine to mimic human naive psychology, that faculty which predisposes us to anthropomorphism and enables us to ascribe intelligence to others. This suggestion would be valid if the test were viewed as a single court case, but not when it is seen as a replicable experiment.

Keywords

False belief tests, folk psychology, naive psychology, the "other minds" problem, theory of mind, the Turing test.
1. The Turing Test allows testers to ask the machine any questions they want, including whether anyone or anything else is intelligent. Watt's (1996) Inverted Test involves asking the machine which of a human-machine pair is intelligent. It is therefore a subset of the Turing Test.

2. It would be possible to program a machine to mimic human behaviour with respect to Naive Psychology alone. Such a machine would pass the Inverted Test but fail the Turing Test.

3. If the Turing Test is considered a single court case lasting five minutes only, the Inverted Test might indeed be the best use of that time, as Naive Psychology may prove the hardest human faculty to mimic. In this case, Watt's argument that the Inverted Test reduces anthropomorphic bias in the original Turing Test would be valid.

4. However, if a machine were ever in practise to pass the Turing Test, a single failiure to replicate this finding would lead to rejecting the hypothesis that the machine was intelligent. As Watt acknowledges, we are as reluctant to ascribe intelligence to a man-made object as we are naturally keen to ascribe it to another human being.

5. Hence the Turing Test for a machine is better viewed as a scientific experiment requiring replication, rather than as a single court case which needs legal justification before it is repeated. As machines which would fail the Inverted Test form only a subset of machines which would fail the Turing Test, the latter is, contrary to Watt's assertions, the most exclusive test.

6. In addition, while alien intelligence is unlikely to be sought in man-made objects, it is possible that the intelligence of something believed by the tester to have an alien origin might be judged with emphasis on aspects of intelligence other than Naive Psychology.

REFERENCES

Watt, S. (1996) Naive Psychology and the Inverted Turing Test. PSYCOLOQUY 7(14) turing-test.1.watt


Volume: 7 (next, prev) Issue: 15 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary
Topic:
Article: