English mathematician Alan Turing (1912-1954) has made many important contributions to mathematics and logic, and is considered one of the fathers of computers and the father of Artificial Intelligence. He was instrumental in breaking the “Enigma” code of the Nazis during WWII, a feat that allowed the Allies to defend their supply lines across the Atlantic and, ultimately, win the war. Two of Turing’s most important contributions to computing are his seminal paper on computable numbers and his paper on the “Imitation Game”.
In the former Turing expands on Gödel’s incompleteness theorem; by assuming a logical machine Turing proved that there is no systematic way of knowing in advance whether such a machine could prove something (a mathematical theorem) to be true or not.
In the latter, he envisages a game whereby a human interrogator queries a “person” without knowing a priori if the person is a man or a woman. Turing showed that the only conclusion that the interrogator could make with regards to the person’s sex would be via his/her answers. Similarly, by replacing the “person” with an intelligent machine Turing argued that if the interrogator could not tell by the answers he got whether the “person” was human or mechanistic then the machine must be regarded as “truly intelligent”.
One of the main criticisms of the Imitation Game – or the “Turing Test”, as it is more commonly known – is its anthropocentricity. Indeed modern AI has moved away from definitions of intelligence specific to humankind, towards a more general – or generic – definition of intelligence. Redefining “intelligent machines” as “intelligent agents” (a more “software” definition) modern AI aims for agents that act so to maximize the expected value of a performance measure, based on past experience and knowledge. Implicit to such a definition are learning and knowledge at the service of some goal-driven operation.
In a way this more “general” definition of artificial intelligence seems like an engineer’s headache pill. By removing such nuances such as behavior and consciousness one is left with little more than a sophisticated control system. Nevertheless, minimalizing the concept of intelligence means that a thermostat must be considered intelligent too – in a similar way that we suspect a nematode worm to have some rudimentary intelligence.
On the contrary, Turing’s Imitation Game challenges us into thinking more deeply about our own consciousness, as well as our perception of others in a social context. This is a much more demanding, richer, and perhaps more annoying problem where engineers – allergic as they might be to philosophers and social scientists meddling into their business – must find the courage to do so in order to solve.



Pingback: Sex with robots III: loving the mecha | Turing Dreams
Pingback: Alan Turing: Celebrating 100 years from his birth | Turing Dreams