Computing Machinery and Intelligence - Allan Turing

Computing Machinery and Intelligence

Allan Turing

Alan Turing, another scientist who rose to prominence during the Second World War (and met a tragic demise), attempts to bring focus to the question of "can machines think?" in his 1950 essay published in Mind magazine. Rather than attempt to answer the question directly (as he admits is an unrealistic task), he instead asks whether a machine could find a solution to a theoretical problem, that a human might have equal or greater difficulty with. Additionally, he describes for the first time a digital computer that follows a program, which reads and writes in alphabetic symbols rather than just numbers. As a demonstration of this he separately developed the Turing Machine.

To set up what he admits is a weak theoretical argument he first works to disarm criticism that he would have faced in his day, regarding the limitations of a discrete-state machines (finite discrete state machines's FSMs- are machines such as digital computers which operate on a principal that they can only be in one state at a time [or access one bit of memory at a time] but can contain capacity for unlimited states or unlimited programming potential, this means that they are universal and can be designed to carry out any pre-programmed function) .

The first two challenges are insignificant, dealing concerns related to theology, and moral consequences. The third anticipated challenge to the notion that "machines can think" is on mathematical grounds, that machines do not possess the same technical capacities of humans. I think neither Turing or his challengers are in a position to decide the question on these grounds because neither at the time knew what the pace of computer capacity would eventually be or exactly what the capacity of the mind is. We still can't exactly quantify how fast the brain works in terms of Hertz or its storage capacity in terms of bites. That machines lack Consciousness is he forth argument he works to disprove. In response he states, "This argument appears to be a denial of the validity of our test. He doesn't believe that the mystery of consciousness must be solved in order to for a mechanical or human entity to solve his problem. The fifth argument is against the machine's general technical limitations. One of these limitation Turing describes is be unrelentingly too perfect. But he counters that a machine could be programed to randomly give the wrong answer to a question as to mask its true nature as a machine. The sixth argument against the machine's ability to meet his test comes from Lady Lovelace a contemporary of Charles Babbage, the inventor of the first theoretical computer. Her argument is that "the Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform." (her emphasis)

In other words, "The machine can never take us by surprise।" While this is true, Turing argues, machines do surprise us not from their own creativity but from our own inabilities to understand what they are doing. The seventh argument revolves around the comparison of the central nervous system between man and machine. Turning admits, despite the general inclination to compare the electronic device's internal circuitry to the human nervous system, that they are completely different - one a discrete state machine and the other a "continuous machine". (Side note: how the theoretical quantum computer affects this argument I don't know. Quantum computers are supposed to be able to operate in multiple states at once, as a opposed to a digital I/O, on or off states of digital machines.) Turning counters that the nature of the nervous system is inconsequential to meet his test. Th eighth argument which Turning anticipates concerning the ability of machines to adapt to every eventuality. But in the case of the Q&A test it is not necessary because the answers are programmed. The final one deals with extra-sensory perception. It baffles me that in the Western Hemisphere in the mid-20th century he would feel the need to address this.

Finally Turing pivots to his own argument for the machine being "intelligent", he basically defers by saying in effect, "what really is intelligence?". He breaks no real ground on a solution other than giving credit to the legitimacy of Lady Lovelace's argument and saying that we'll have to wait until the end of century to find out the true answer. All honest statements.

He cites the future need of programming capacity over hardware capacity. The reason being is that once again we are seeing how futurists of the mid-20th century are already predicting the advances in hardware technology of the new millennium.

To program the intelligent computers to match the human brain, Turing suggests starting with aiming program at mimicking a child's brain. This way he argues, "If this then were then subjected to an appropriate course of education one would obtain the adult brain." In terms of practically building an AI system this method seems a little presumptuous. But, I think he's on the right track with trying to build human intelligence or "artificial intelligence" from a smaller scale and then trying to grow it into something. Not only do I think it would be more feasible to start small but I think that the growing element is maybe the most important indicator of intelligence.

For more background on the evolution of the field of artificial intelligence, check out this Wired Magazine retrospective