Rethinking turing's test

Journal of Philosophy

Volumn 110, Issue 7, 2013, Pages 391-411

  Proudfoot, Diane ^a  

^a UNIVERSITY OF CANTERBURY   (New Zealand)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 84890032410     PISSN: 0022362X     EISSN: 19398549     Source Type: Journal    
DOI: 10.5840/jphil2013110722     Document Type: Article

References (98)

1. On the Manchester computers, see further B. Jack Copeland and Diane Proudfoot, "Artificial Intelligence: History, Foundations, and Philosophical Issues," in Paul Thagard, ed., Handbook of the Philosophy of Science: Philosophy of Psychology and Cognitive Science (Amsterdam: Elsevier, 2006), pp. 429-82;
2. and Copeland and Proudfoot, "Turing and the Computer," in Copeland et al., Alan Turing's Electronic Brain: The Struggle to Build the ACE, the World's Fastest Computer (New York: Oxford, 2012), pp. 107-48.
3. For a comprehensive history, see Copeland, "The Manchester Computer: A Revised History," IEEE Annals of the History of Computing, xxxiii, 1 ( January-March 2011): 4-37.
4. Alan M. Turing, "Computing Machinery and Intelligence," Mind, lix, 236 (October 1950): 433-60. See p. 436. turing 391 0022-362X/13/1007/391- 411
5. Turing, "Intelligent Machinery," National Physical Laboratory Report (1948).
6. Reproduced in Copeland, ed., The Essential Turing (New York: Oxford, 2004), pp. 410-32.
7. Reproduced in Copeland et al., Alan Turing's Electronic Brain, pp. 369-454. See p. 389.
8. Turing, "Intelligent Machinery," pp. 431, 412.
9. Turing, Richard Braithwaite, Geoffrey Jefferson, and Max Newman, "Can Automatic Calculating Machines Be Said to Think.," BBC radio broadcast (1952).
10. Copeland and Proudfoot, "Artificial Intelligence: History, Foundations, and Philosophical Issues" and "Turing and the Computer."
11. Turing et al., "Can Automatic Calculating Machines Be Said to Think ," p. 495. In 1950 Turing famously predicted that "in about fifty years' time it will be possible to programme computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 per cent. chance of making the right identification after five minutes of questioning"
12. ("Computing Machinery and Intelligence," p. 442). His 1952 prediction does not specify the computer's storage capacity or the test's duration. In response to Max Newman's question as to when a machine will "stand any chance [of passing the test] with no questions barred," Turing simply said "at least 100 years"
13. ("Can Automatic Calculating Machines Be Said to Think ," p. 495).
14. Turing, "Computing Machinery and Intelligence," p. 442.
15. Wolfe Mays, "Can Machines Think ," Philosophy, xxvii, 101 (April 1952): 148-62.
16. See pp. 151, 160. A recent example of the canonical view is Paul R. Cohen, "If Not Turing's Test, Then What ," AI Magazine, xxvi, 4 (Winter 2005): 61-67.
17. James H. Moor, "An Analysis of the Turing Test," Philosophical Studies, xxx, 4 (October 1976): 249-57;
18. and Moor, "Turing Test," in Stuart Shapiro, ed., Encyclopedia of Artificial Intelligence, vol. 2 (New York: Wiley, 1987), pp. 1126-30.
19. Copeland, Artificial Intelligence: A Philosophical Introduction (Oxford: Blackwell, 1993);
20. and Copeland, "The Turing Test," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence (Boston: Kluwer, 2003), pp. 1-21.
21. Copeland quotes this in "The Turing Test," p. 5.
22. Early proponents of the canonical view did not distinguish between an operationalist and a dispositional (or capacity) interpretation of the Turing test, unlike later proponents (such as Ned Block, "Psychologism and Behaviorism," Philosophical Review, xc, 1 ( January 1981): 5-43;
23. and Block, "The Mind as the Software of the Brain," in Edward E. Smith and Daniel N. Osherson, eds., An Invitation to Cognitive Science, Volume 3: Thinking, 2nd ed. (Cambridge: MIT, 1995), pp. 377-425)
24. A recent example of the popular variant of the canonical view is Eugene Eberbach, Dina Goldin, and Peter Wegner, "Turing's Ideas and Models of Computation," in Christof Teuscher, ed., Alan Turing: Life and Legacy of a Great Thinker (New York: Springer, 2004), pp. 159-94.
25. See Proudfoot, "A New Interpretation of the Turing Test," The Rutherford Journal: The New Zealand Journal for the History and Philosophy of Science and Technology, i (2005-2006). rutherfordjournal.org.
26. Andrew Hodges, Alan Turing: The Enigma (London: Vintage, 1992), p. 415.
27. Commentators who take this line include Judith Genova, "Turing's Sexual Guessing Game," Social Epistemology, viii, 4 (1994): 313-26;
28. Patrick Hayes and Kenneth Ford, "Turing Test Considered Harmful," IJCAI-95 Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence, Montreal, Quebec, August 20-25, 1995, vol. I (Morgan Kaufman, 1995), pp. 972-77;
29. Susan G. Sterrett, "Turing's Two Tests for Intelligence," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence, pp. 79-97;
30. Douglas B. Lenat, "The Voice of the Turtle: Whatever Happened to AI ," AI Magazine, xxix, 2 (Summer 2008): 11-22;
31. Lenat, "Building a Machine Smart Enough to Pass the Turing Test: Could We, Should We, Will We ," in Robert Epstein, Gary Roberts, and Grace Beber, eds., Parsing the Turing Test: Philosophical and Methodological Issues in the Quest for the Thinking Computer (Berlin: Springer, 2008), pp. 261-82.
32. Copeland points this out ("The Turing Test," p. 9).
33. Ayse Pinar Saygin, Ilyas Cicekli, and Varol Akman, "Turing Test: 50 Years Later," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence, pp. 23-78. See pp. 26, 69.
34. I point this out in a preliminary way in Proudfoot, "A New Interpretation of the Turing Test."
35. On response-dependent concepts, see, for example, Mark Johnston, "Dispositional Theories of Value," Proceedings of the Aristotelian Society, Supplementary Volumes, lxiii (1989): 139-74;
36. and Philip Pettit, "Realism and Response-Dependence," Mind, c, 4 (October 1991): 587-626.
37. Johnston, "Dispositional Theories of Value," p. 145.
38. Recent accounts include Jussi Haukioja, "How (Not) to Specify Normal Conditions for Response-Dependent Concepts," Australasian Journal of Philosophy, lxxxv, 2 ( June 2007): 325-31;
39. and David Yates, "Response-Dependence," Philosophical Books, xlix, 4 (October 2008): 344-54.
40. According to Block, "there is a gap in Turing's proposal: we are not told how the judge is to be chosen. A judge who was a leading authority on genuinely intelligent machines might know how to tell them apart from people....Turing acknowledged this point by jettisoning the claim that being able to pass the Turing test is a necessary condition of intelligence" ("The Mind as the Software of the Brain," pp. 378-79). However, Turing did exclude experts-and at no point did he suggest that the game provided a necessary condition of intelligence (see section i).
41. See Pettit, "Realism and Response-Dependence."
42. Moor, "An Analysis of the Turing Test," p. 250.
43. Kenneth Ford and Patrick Hayes, "On Conceptual Wings: Rethinking the Goals of Artificial Intelligence," Scientific American Presents, ix (1998): 78-83. See p. 81.
44. Paul Churchland, "Learning and Conceptual Change: The View from the Neurons," in Andy Clark and Peter Millican, eds., The Legacy of Alan Turing, Volume II: Connectionism, Concepts, and Folk Psychology (Oxford: University Press, 1996), pp. 7-43. See p. 8.
45. Block, "The Mind as the Software of the Brain," p. 378.
46. Block claims that Turing "was willing to settle for a 'sufficient condition' formulation of his behaviorist definition of intelligence" ("Psychologism and Behaviorism," p. 15). See note 32.
47. Block, "Psychologism and Behaviorism," p. 18.
48. See John N. Crossley and Lloyd Humberstone, "The Logic of 'Actually'," Reports on Mathematical Logic, viii (1977): 11-29.
49. See Martin Davies and Humberstone, "Two Notions of Necessity," Philosophical Studies, xxxviii, 1 ( July 1980): 1-30.
50. See pp. 21-24. In Crossley and Humberstone's original paper, they suggest that "actually" has the logic of a "rigid operator," using the phrase analogously to Kripke's "rigid designator" ("The Logic of 'Actually'," p. 14).
51. See, for example, Johnston, "Dispositional Theories of Value"; and Yates, "Response-Dependence."
52. On rigid and non-rigid response-dependence theories see, for example, Nick Zangwill, "Skin Deep or in the Eye of the Beholder. The Metaphysics of Aesthetic and Sensory Properties," Philosophy and Phenomenological Research, lxi, 3 (November 2000): 595-618
53. Mark LeBar, "Three Dogmas of Response-Dependence," Philosophical Studies, cxxiii, 3 (April 2005): 175-211.
54. Moor, "An Analysis of the Turing Test"; "Turing Test"; and "Explaining Computer Behavior," Philosophical Studies, xxxiv, 3 (October 1978): 325-27.
55. Moor, "The Status and Future of the Turing Test," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence, pp. 197-213. See p. 202.
56. "The Status and Future of the Turing Test," pp. 80-81.
57. see also "Turing Test," p. 1127.
58. Stuart Shieber, "The Turing Test as Interactive Proof," Noûs, xli, 4 (December 2007): 686-713. To avoid confusion, it is worth pointing out that, in Shieber's sense, "interaction" is quite unlike Turing's requirement that the interrogator be "taken in" by the machine; rather, the interrogator and the contestant "engage in rounds of messagepassing" (ibid., p. 696). Likewise, although Eberbach, Goldin, and Wegner use the phrase 'interactive Turing Test', as they use the expression an interactive test is just one where the computer interacts with the environment-that is, the human interrogator ("Turing's Ideas and Models of Computation," p. 185). This is again very different from Turing's emphasis on the interrogator's response (Eberbach, Goldin, and Wegner take the canonical view that the Turing test is behaviorist).
59. Shieber, "The Turing Test as Interactive Proof," p. 706. Shieber does not argue that the modified compact conception is true, but only that Block's putative counterexample does not defeat this conception.
60. see also Pettit, "A Theory of Normal and Ideal Conditions," Philosophical Studies, xcvi, 1 (October 1999): 21-44.
61. Pettit, "A Theory of Normal and Ideal Conditions," pp. 592, 611-15.
62. Piccinini, "Turing's Rules for the Imitation Game," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence, pp. 111-20. See p. 117
63. Piccinini remarks: Turing called intelligence an "emotional concept," meaning that there is no objective way to apply it. Direct experience with a machine, which plays chess in a way that cannot be distinguished from human playing, could convince one to attribute intelligence to the machine. This is very likely to be an important part of the historical root for Turing's proposal of the imitation game....[Turing] hoped that, by experiencing the versatility of digital computers at tasks normally thought to require intelligence, people would modify their usage of terms like "intelligence" and "thinking," so that such terms apply to the machines themselves. ("Turing's Rules for the Imitation Game," p. 117)
64. On this reading, Turing's goal for the imitation game is pragmatic-altering common linguistic usage. However, Turing claimed that the imitation game provides a criterion of thinking. Although he predicted that linguistic usage would change ("Computing Machinery and Intellligence," p. 442)
65. Turing, "Can Digital Computers Think ," p. 486.
66. Moor, "The Status and Future of the Turing Test," p. 203.
67. Rodney A. Brooks, "Intelligence without Reason," in Luc Steels and Brooks, eds., The Artificial Life Route to Artificial Intelligence (Hillsdale, NJ: Lawrence Erlbaum, 1995), pp. 25-81. See p. 57.
68. Jordan B. Pollack, "Mindless Intelligence," IEEE Intelligent Systems, xxi, 3 (May/June 2006): 50-56. See p. 51.
69. See Proudfoot, "Anthropomorphism and AI: Turing's Much Misunderstood Imitation Game," Artificial Intelligence, clxxv, 5-6 (April 2011): 950-57;
70. and Proudfoot, "The Implications of an Externalist Theory of Rule-Following Behaviour for Robot Cognition," Minds and Machines, xiv, 3 (August 2004): 283-308.
71. See Cynthia Breazeal, Designing Sociable Robots (Cambridge: MIT, 2002)
72. Breazeal, "Emotive Qualities in Lip-Synchronized Robot Speech," Advanced Robotics, xvii, 2 (2003): 97-113;
73. Breazeal, "Role of Expressive Behaviour for Robots that Learn from People," Philosophical Transactions of the Royal Society B, ccclxiv, 1535 (Dec. 12, 2009): 3527-38
74. AI researchers deliberately exploit the tendency to anthropomorphize, for example to aid human-computer interaction (on the efficacy of anthropomorphism, see Li Gong, "How Social is Social Responses to Computers The Function of the Degree of Anthropomorphism in Computer Representations," Computers in Human Behavior, xxiv, 4 ( July 2008): 1494-509)
75. Kismet is constructed so that untrained human observers believe that they understand its "facial" and "bodily" displays, and respond to its behavior as to ordinary human social signals (see Breazeal, "Toward Sociable Robots," Robotics and Autonomous Systems, xlii, 3-4 (Mar. 31, 2003): 167-75;
76. and Breazeal, "Emotion and Sociable Humanoid Robots," International Journal of Human-Computer Studies, lix, 1-2 ( July 2003): 119-55)
77. The grand goal is to build a "socially intelligent" robot (see Kerstin Dautenhahn, "Socially Intelligent Robots: Dimensions of Human-Robot Interaction," Philosophical Transactions of the Royal Society B, ccclxii, 1480 (2007): 679-704)
78. Breazeal and Paul Fitzpatrick, "That Certain Look: Social Amplification of Animate Vision," in Proceedings of the AAAI Fall Symposium, Socially Intelligent Agents: The Human in the Loop (2000), http://people.csail.mit.edu/paulfitz/publications.shtml
79. Breazeal, "Affective Interaction between Humans and Robots," in Jozef Kelemen and Petr Sosík, eds., Advances in Artificial Life: 6th European Conference, ECAL 2001 (Berlin: Springer-Verlag, 2001), pp. 582-91.
80. Breazeal and Juan Velásquez, "Toward Teaching a Robot 'Infant' using Emotive Communication Acts," in Proceedings of 1998 Simulation of Adaptive Behavior, Workshop on Socially Situated Intelligence (1998). http://www.ai.mit.edu/projects/sociable/publications.html.
81. Breazeal, "Affective Interaction between Humans and Robots," p. 585
82. "Toward Sociable Robots," p. 147.
83. Breazeal and Fitzpatrick, "That Certain Look: Social Amplification of Animate Vision"
84. Breazeal, "Toward Sociable Robots," p. 172
85. On "expressive" face robots, see Proudfoot, "Can a Robot Smile Wittgenstein on Facial Expression," forthcoming in Timothy P. Racine and Kathleen L. Slaney, eds., A Wittgensteinian Perspective on the Use of Conceptual Analysis in Psychology (New York: Palgrave Macmillan, 2013), pp. 172-84.
86. I discuss this problem with respect to current developments in AI in Proudfoot, "Anthropomorphism and AI: Turing's Much Misunderstood Imitation Game."
87. http://www.loebner.net/Prizef/loebner-prize.html.
88. For example, a judge in the 2003 competition remarked "I am not to[o] keen on people including what I have just said in their replies." http://www.loebner.net/Prizef/loebner-prize.html.
89. Some AI researchers claim that the Turing test encourages anthropomorphism. For example, Brian Duffy says of the test, "[H]ow can one... not succumb to such tricks and degrees of anthropomorphising...."
90. (Duffy, "Anthropomorphism and the Social Robot," Robotics and Autonomous Systems, xlii, 3-4 (Mar. 31, 2003): 177-90; see p. 179). However, the behavior of Loebner Contest judges shows the contrary: the design of Turing's game discourages anthropomorphism.
91. (Loebner, "How to Hold a Turing Test Contest," in Epstein, Roberts, and Beber, eds., Parsing the Turing Test: Philosophical and Methodological Issues in the Quest for the Thinking Computer, pp. 173-79).
92. As described in Turing et al., "Can Automatic Calculating Machines Be Said to Think.". Copeland ("The Turing Test") points this out.
93. Recent examples of this claim are found in William J. Rapaport, "How to Pass a Turing Test," in Moor, ed., The Turing Test: The Elusive Standard of Artificial Intelligence, pp. 161-84
94. Robert M. French, "The Turing Test: The First 50 Years," Trends in Cognitive Science, iv, 3 (Mar. 1, 2000): 115-22. See p. 116.
95. Linnda Caporael argues that Turing's imitation game cannot be used as a criterion for thinking; she says, "[I]f there is an inherent anthropomorphic bias [in human reasoning], the imitation principle clearly incorporates a confirmatory bias, and is therefore unreliable" (Caporael, "Anthropomorphism and Mechanomorphism:Two Faces of the Human Machine," Computers in Human Behavior, ii, 3 (1986): 215-34; see pp. 228-29). On the contrary, any "inherent anthropomorphic bias" is a controlled variable in Turing's experiment.
96. French, "Subcognition and the Limits of the Turing Test," in Millican and Clark, eds., The Legacy of Alan Turing, Volume I: Machines and Thought (Oxford: University Press, 1996), pp. 11-26. See p. 12.
97. Turing, "Intelligent Machinery, A Heretical Theory," BBC radio broadcast (c. 1951). Reproduced in Copeland, ed., The Essential Turing, pp. 472-75;
98. see p. 473. Likewise, in Turing's 1952 radio discussion, Richard Braithwaite remarked that "to protect a machine against a too hostile environment it may be essential to allow it, as it were, to go to bed with a neurosis, or psychogenic illness"