Abstract
Alan Turing’s 1950 imitation game has been widely understood as a means for testing if an entity is intelligent. Following a series of papers by Diane Proudfoot, I offer a socio-technological interpretation of Turing’s paper and present an alternative way of understanding both the imitation game and Turing’s concept of intelligence. Turing, I claim, saw intelligence as a social concept, meaning that possession of intelligence is a property determined by society’s attitude toward the entity. He realized that as long as human society held a prejudiced attitude toward machinery—seeing machines a priori as mindless objects—machines could not be said to be intelligent, by definition. He also realized, though, that if humans’ a priori, chauvinistic attitude toward machinery changed, the existence of intelligent machines would become logically possible. Turing thought that such a change would eventually occur: He believed that when scientists overcome the technological challenge of constructing sophisticated machines that could imitate human verbal behavior—i.e., do well in the imitation game—humans’ prejudiced attitude toward machinery will have altered in such a way that machines could be said to be intelligent. The imitation game, for Turing, was not an intelligence test, but a technological aspiration whose realization would likely involve a change in society’s attitude toward machines.
Similar content being viewed by others
Availability of Data and Material
Not applicable.
Code Availability
Not applicable.
Notes
Turing uses the terms “thinking (entity)” and “intelligent (entity)” interchangeably, as Piccinini (2000) and others have pointed out. I will not differentiate between the terms, although I will usually use the term “intelligence”.
Other inductive interpretations include Watt (1996), Schweizer (1998), and Shieber (2007). Sheiber suggests seeing the imitation game as an interactive proof: He shows that under any reasonable statistical measure, the chance of a non-intelligent entity faking intelligible answers over several rounds of the imitation game is negligible. Gonçalves (2021), too, might be considered as promoting an inductive interpretation, as he sees the imitation game as a means of discovering if a machine possesses the non-observable property of “the ability to learn from one’s own experience”.
Turing attended Wittgenstein’s lectures in Cambridge on the foundations of mathematics. Sadly, there are no records of them discussing issues directly related to machine intelligence. Nonetheless, I think Turing’s approach to intelligence is quite Wittgensteinian, as will soon be shown.
Wittgenstein himself seems to hold that the possession of any mental property by an entity (e.g., consciousness, agency, free will) is always “from a certain perspective.” I think Turing would agree with such a generalization, but I will limit my discussion to Turing’s approach to intelligence alone. (Cf. Proudfoot, 2017 and 2020), who argues that Turing held a “response-dependent” approach to both intelligence and free will, and perhaps also to consciousness).
This is a suggestion I made in Danziger (2018).
Turing’s approach as described in this section bears resemblance to Dennett’s “intentional stance” (Dennett, 1987a).
I thank an anonymous reviewer for helpful comments regarding this section.
Proudfoot’s formulation of Turing’s view of intelligence in terms of response-dependency includes explicit reference to the imitation game. My interpretation of Turing differs from Proudfoot’s in some essential points, including in the way intelligence is formalized as a response-dependent property; I present my view here and discuss Proudfoot’s interpretation later (section 4.5).
Brynjarsdóttir (2008) shows that although Johnston, who had coined the term “response-dependence,” had originally mentioned “response-dependent concepts” (a term which, taken literally, is to be understood as referring to cases in which one’s conceptualization of a property depends on some subjective response), his account actually describes response-dependent properties (cases in which the property itself is ontologically dependent on some subjective response). There have been interesting attempts to formulate a response-dependent account of concepts (Pettit, (1991), Pettit, (1998), Jackson & Pettit, (2002)); according to the interpretation presented in this paper, though, it seems more suitable to say that Turing saw intelligence possession as a response-dependent property.
Other formulations for response-dependent properties have been suggested. See Yates (2008) for an example of one such formulation that specifies the requirements of a prioricity (of the biconditional) and substantiality (of the terms specifying the conditions K).
This is how Yates (2008) introduces the idea of a response-dependent property.
In this paper, “displaying intelligent-like behavior” and “behaving in an intelligent-like manner” should be understood as “displaying behavior that under regular circumstances cannot be differentiated from that of a human”.
Turing’s apparent agreement here with (what I called) “Wittgenstein’s iron curtain of language conventions” is the basis for the claim I made in section 3.1 above, namely that Turing, in his 1950 paper, did not see intelligence as defined by the viewpoint of some individual observer or another (as may have been his view in his earlier publications) but as defined by the attitude of society. Not unrelated, in Danziger (2018) I suggested that Turing’s refraining from explicitly stating in his 1950 paper that machines could think might imply that he had taken a step back from the stance expressed in his earlier writings.
Turing’s sociological prediction has been discussed by Mays (1952, 149–151), Rapaport (2000), Beran (2014), and Gonçalves (2021). Rapaport (2000) suggests an interesting differentiation between two kinds of possible socio-linguistic changes: One consists of an extension of the scope of terms like “intelligence” to cover machines, similar to the extension of the scope of the term “flying” to cover movement-through-the-air of various aircraft, and not just of birds. Such an extension, says Rapaport, may be metaphorical; it does not necessarily imply that language users perceive the flight of airplanes (or machine intelligence) the same way they perceive the flight of birds (or human intelligence). Another possible change of the “general educated opinion,” according to Rapaport, is one in which people would come to see machines as possessing real intelligence. Albeit the fact that Turing described the change in humans’ attitude as a linguistic alteration, I think he had in mind Rapaport’s second type of change.
Cf. Sloman (2013), who also claims that Turing saw the imitation game as a technological challenge.
See Yampolskiy (2013) for a literature review and a more formal account of AI-completeness, including explicit reference to the imitation game.
[Author's note: A “paper machine” was a person whose role was to execute a given algorithm step by step, in a fully mechanical manner; this was how a program’s functionality was tested before digital computers were available for use].
As mentioned, the passages quoted from Turing which I used to exemplify his descriptive manner were taken from his 1947 and 1948 papers. It should be noted that there seems to be a slight difference between the abilities of the digital computer emphasized in these papers and the ability highlighted in Turing’s 1950 paper. In the 1947 and 1948 papers, Turing stresses that digital computers could display those specific, unique abilities that he sees as the hallmarks of intelligence, such as the ability to learn from one’s mistakes and the ability to modify one’s own program (“instruction table”) during runtime. In the 1950 paper, it seems that Turing is no longer trying to convince the reader that digital computers could imitate some specific cognitive ability or another; instead, he stresses that digital computers could imitate the entire human cognitive system, as they could imitate the entire human brain. (Hodges (2014, 530) explains the difference between Turing’s1948 and 1950 papers in a similar way. See Danziger (2018) for a detailed comparison between the approaches expressed in Turing’s 1947, 1948, and 1950 papers.) Despite this possible difference, in all three papers Turing describes the way humans would react (or actually do react) upon encountering such sophisticated machines, and the way humans’ attitude toward machinery would be influenced; he does not declare, though, that one should react this way or another, or that humans’ attitude toward machinery should change in some way or another, upon encountering sophisticated machines. This commonality between Turing’s papers is what allows me to explicate the descriptive manner in his 1950 paper by quoting from his 1947 and 1948 papers, where his descriptive intentions are more telling.
Steven Harnad (in Epstein et al., 2009, 48) remarks that what Turing calls here “solipsism” is actually the “other-minds” problem in philosophy.
Indeed, it seems that Turing wanted humans’ attitude toward machinery to change. Robin Gandy, who had been Turing’s student and close friend, says that Turing sought to persuade people that “computers were not merely calculating engines but were capable of behaviour which must be accounted as intelligent” (Gandy, 1996, 125). Likewise, Piccinini (2000), who holds a similar understanding of the role of the imitation game as the one presented in this paper, suggests that 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” (Piccinini, 2000), 579). See Gonçalves (2021, chap. 1) for an analysis of Turing’s ambition(s); he concludes that Turing’s 1950 paper can be seen as the point in time in which Turing assumed his role as “prophet of the machines”.
Cf. Gonçalves (2021, chap. 3), who claims that Turing’s imitation game should be seen as a thought experiment in science, as opposed to scholars who see it as a thought experiment in philosophy.
Note, though, that things may be different with regard to Turing’s 1947 and 1948 papers. There, Turing may have actually been focusing on the intellectual status of the machine itself (see Danziger, 2018).
See Proudfoot (2020, fn. 1) for interesting remarks by Marvin Minsky and Drew McDermott regarding this issue.
This is in contrast with to my own formulation of Turing’s response-dependent definition of intelligence in section 3.2 above, which includes no reference to the game.
Bringsjord et al. (2001) suggest a criterion for intelligence called the restricted epistemic relation, which is similar to what Abramson saw as Turing’s epistemic-limitation condition, except that Bringsjord et al. do not think that Turing himself required this criterion. Accordingly, they suggest the Lovelace test for intelligence, by which a necessary condition for an entity’s being considered intelligent is that its creator does not know how it produces its answer. Abramson thinks that this is how Turing himself saw the imitation game.
It may be noted that in the 1952 radio broadcast mentioned in section 3.4, Turing said that development of machines that do well in the imitation game would take “at least 100 years” (Turing et al., 1952, 495). Possibly, Turing realized that technological development of digital computers was progressing slower than he had expected, and so he updated the timeframe of his prediction accordingly, leaving it open-ended, with no deadline.
Sloman (2013, 608) makes a similar point, and points out that while computers are now doing much cleverer things than in the past, people are becoming much harder to impress.
While Turing regarded the entity’s external appearance as insignificant for intelligence attribution (Turing, 1950, 434), others have argued that external appearance may indeed have an impact on society’s attitude toward the entity, for better or worse. Moreover, such an impact is not necessarily “linear”: Under certain circumstances, cases of similarity between humans and other entities might actually sharpen the difference between them. For example, Mori’s “Uncanny Valley” graph (Mori et al., 2012) claims that an entity that is very similar to humans in one aspect but quite different from them in another, might be perceived by humans as very un-human.
Cf. Davidson (1990), who claims that Turing had seen this property as irrelevant for intelligence attribution.
For a list of several other properties that may trigger intelligence attribution, see Torrance (2014, 25–26).
See Rapaport (2000) for a discussion of cases in which there is a controversy between a subject and society regarding the subject’s mental properties.
References
Abramson, D. (2008). Turing’s responses to two objections. Minds and Machines, 18, 147–167.
Abramson, D. (2011). Descartes’ influence on Turing. Studies in History and Philosophy of Science, 42, 544–551.
Beran, O. (2014). Wittgensteinian Perspectives on the Turing Test. Studia Philosophica Estonica 7(1), 35–57.
Block, N. (1981). Psychologism and behaviorism. Philosophical Review, 90(1), 5–43.
Block, N. (1995). The mind is the software of the brain. In E. E. Smith & D. N. Osherson (Eds.), Thinking (pp. 377–425). MIT Press.
Boden, M. A. (2006). Mind as machine: A history of cognitive science. Oxford University Press.
Bringsjord, S., Bello, P., & Ferrucci, D. (2001). Creativity, the Turing test, and the (better) Lovelace test. Minds and Machines, 11, 3–27.
Brynjarsdóttir. (2008). Response-dependence of concepts is not for properties. American Philosophical Quarterly, 45(4), 377–386.
Chomsky, N. (2009). Turing on the “imitation game.” In Epstein et al. (pp. 103–106).
Copeland, B. J. (Ed.). (2004). The essential Turing. Oxford University Press.
Davidson, D. (1990). Turing’s test. In K. Said, W. Newton-Smith, R. Viale, & K. Wilkes (Eds.), Modelling the mind (pp. 1–12). Clarendon Press.
Danziger, S. (2018). Where intelligence lies: Externalist and sociolinguistic perspectives on the Turing test and AI. In V.C. Müller (Ed.), Philosophy and theory of artificial intelligence 2017. Studies in Applied Philosophy, Epistemology and Rational Ethics, vol. 44 (pp. 158–174). Springer.
Dennett, D. C. (1987a). The intentional stance. MIT Press.
Dennett, D. C. (1987b). Consciousness. In R. L. Gregory & O. L. Zangwill (Eds.), The Oxford companion to the mind (pp. 160–164). Oxford University Press.
Epstein, R., Roberts, G., & Beber, G. (Eds.). (2009). Parsing the Turing test: Philosophical and methodological issues in the quest for the thinking computer. Springer.
French, R. M. (1990). Subcognition and the limits of the Turing test. Mind, 99(393), 53–65.
Gandy, R. (1996). Human versus mechanical intelligence. In P. J. R. Millican & A. Clark (Eds.), Machines and thought: The legacy of Alan Turing (Vol. 1, pp. 125–136). Oxford University Press.
Gonçalves, B. (2021). Machines will think: Structure and interpretation of Alan Turing’s imitation game. Doctoral Dissertation, Faculty of Philosophy, Languages and Human Sciences, University of São Paulo, São Paulo. https://doi.org/10.11606/T.8.2021.tde-10062021-173217. Accessed 19 Apr 2022.
Harnad, S. (1991). Other bodies, other minds: A machine incarnation of an old philosophical problem. Minds and Machines, 1, 43–54.
Hodges, A. (2014). Alan Turing: The enigma. Princeton University Press.
Jackson, F., & Pettit, P. (2002). Response-dependence without tears. Philosophical Issues, 12, 96–117.
Johnston, M. (1989). Dispositional theories of value. Proceedings of the Aristotelian Society, Supplementary Volumes, 63, 139–174.
Mays, W. (1952). Can machines think? Philosophy, 27, 148–162.
McCorduck, P. (2004). Machines who think: A personal inquiry into the history and prospects of artificial intelligence. CRC Press.
Michie, D. (1993). Turing’s test and conscious thought. Artificial Intelligence, 60(10), 1–22.
Minsky, M. (1986). The society of mind. Simon & Schuster.
Moor, J. H. (1976). An analysis of the Turing test. Philosophical Studies, 30, 249–257.
Moor, J. H. (2001). The status and future of the Turing test. Minds and Machines, 11, 77–93.
Mori, M., MacDorman, K. F., & Kegeki, N. (2012). The uncanny valley. IEEE Robots & Automation Magazine, 19(2), 98–100.
Oppy, G., & Dowe, D. (2011). The Turing test. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. plato.stanford.edu/archives/spr2011/entries/turing-test. Accessed 13 Oct 2017.
Pettit, P. (1991). Realism and response-dependence. Mind, 100, 587–626.
Pettit, P. (1998). Terms, things and response-dependence. European Review of Philosophy, 3, 61–72.
Piccinini, G. (2000). Turing’s rules for the imitation game. Minds and Machines, 10, 573–582.
Proudfoot, D. (2013). Rethinking Turing’s test. The Journal of Philosophy, 110(7), 391–411.
Proudfoot, D. (2017). Turing and free will: A new take on an old debate. In J. Floyd & A. Bokulich (Eds.), Philosophical explorations of the legacy of Alan Turing (pp. 305–321). Springer Verlag.
Proudfoot, D. (2020). Rethinking Turing’s test and the philosophical implications. Minds and Machines, 30, 487–512.
Proudfoot, D. (2005). A new interpretation of the Turing test. The Rutherford Journal: The New Zealand Journal for the History and Philosophy of Science and Technology,1. rutherfordjournal.org/article010113.html. Accessed 6 Oct 2020.
Rapaport, W. J. (2000). How to pass a Turing test. Journal of Logic, Language, and Information, 9(4), 467–490.
Saygin, A., Cicekli, I., & Akman, V. (2000). Turing test: 50 years later. Minds and Machines, 10, 463–518.
Schweizer, P. (1998). The truly total Turing test. Minds and Machines, 8, 263–272.
Searle, J. R. (1980). Minds, brains, and programs. The Behavioral and Brain Science, 3, 417–424.
Shapiro, S. C. (2003). Artificial intelligence (AI). In A. Ralston, E. D. Reilly, & D. Hemmendinger (Eds.), Encyclopedia of computer science (pp. 89–93). Wiley.
Shieber, S. M. (2007). The Turing test as interactive proof. Noûs, 41(4), 686–713.
Sloman, A. (2013). Aaron Sloman absolves Turing of—The mythical Turing test. In S. B. Cooper & J. Van Leeuwen (Eds.), Alan Turing: His work and impact (pp. 606–611). Elsevier.
Tesler, L. (ca. 1970). Tesler’s theorem. http://www.nomodes.com/Larry_Tesler_Consulting/Adages_and_Coinages.html. Accessed 30 Apr 2020.
Torrance, S. (2014). Artificial consciousness and artificial ethics: Between realism and social relationism. Philosophy and Technology, 27(1), 9–29.
Turing, A. M. (1950). Computing machinery and intelligence. Mind, 50, 433–460.
Turing, A. M., Braithwaite, R., Jefferson, G., & Newman, M. (1952). Can automatic calculating machines be said to think? Reprinted in Copeland (pp. 494–506).
Turing, A. M. (1936). On computable numbers, with an application to the Entscheidungsproblem. Reprinted in Copeland (pp. 58–90).
Turing, A. M. (1947). Lecture on the automatic computing engine. Reprinted in Copeland (pp. 378–394).
Turing, A. M. (1948). Intelligent machinery. Reprinted in Copeland (pp. 410–432).
Watt, S. (1996). Naive psychology and the inverted Turing test. Psycoloquy, 7(14). http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.43.2705&rep=rep1&type=pdf. Accessed 2 Nov 2017.
Whitby, B. (1996). The Turing test: AI’s biggest blind alley? In P. Millican & A. Clark (Eds.), Machines and thought: The legacy of Alan Turing (pp. 53–62). Calderon Press.
Wittgenstein, L. (1958). The blue and brown books: Preliminary studies for the “philosophical investigations.” Harper & Row.
Wittgenstein, L. (1976). Wittgenstein’s lectures on the foundations of mathematics, Cambridge, 1939. Cornell University Press.
Wittgenstein, L. (2017). Lectures on freedom of the will. In V. A. Muntz & B. Ritter (Eds.), Wittgenstein’s Whewell’s court lectures, Cambridge, 1938–1941, from the notes by Yorick Smythies (pp. 282–296). Wiley-Blackwell.
Wittgenstein, L. (2009). Philosophical investigations. Trans: G. E. M. Anscombe, P. M. S. Hacker, & J. Schulte fourth edition. Wiley-Blackwell.
Wright, C. (1992). Truth and objectivity. Harvard University Press.
Yampolskiy, R. V. (2013). Turing test as a defining feature of AI-completeness. In X. S. Yang (Ed.), Artificial intelligence, evolutionary computing and metaheuristics: In the footsteps of Alan Turing. Studies in Computational Intelligence, vol. 427 (pp. 3–17). Springer.
Yates, D. (2008). Philosophical Books, 49(4), 344–354.
Acknowledgements
I thank Orly Shenker, Oron Shagrir, Netanel Kupfer, and two anonymous reviewers, who have read various versions of this paper and added thoughtful comments. Special thanks to Nick Novelli, Nicola Damassino, and the participants of the R3T3-2018 Conference at the University of Edinburgh for stimulating discussions, and to Shira Kramer-Danziger for her assistance in editing and her wise advice.
Funding
Research for this paper was financially supported by the Sidney M. Edelstein Center for the History and Philosophy of Science, Technology, and Medicine at the Hebrew University of Jerusalem; and by the School of Philosophy, Psychology, and Language Sciences at the University of Edinburgh.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing Interests
The author declares no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Danziger, S. Intelligence as a Social Concept: a Socio-Technological Interpretation of the Turing Test. Philos. Technol. 35, 68 (2022). https://doi.org/10.1007/s13347-022-00561-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s13347-022-00561-z