by which the human thinker expresses and externalises their own private thoughts . The computer is unthinking in the same sense as the pen held by the hand is unthinking when it inscribes our inky signs onto a page . And at the same time , whatever underlies our meaningful thought does not appear like it can be entirely captured by reduction down to these symbols and signs , but symbols and signs are all we ever have available to describe and build a thinking machine .
The only extant machines that might reasonably be considered as intelligent — semiconductor devices comprising billions of transistors , arranged to electronically simulate the algorithmic operations of Turing ' s abstract universal machine — are , at their fundamental level , only ever performing symbolic mathematical operations , and so there is no ultimate understanding of the meaning behind their rule-following . Indeed , these rule-following considerations , which a . re a crucial part of thought and understanding , form an important discussion in Wittgenstein ' s Philosophical Investigations , and serve to demonstrate that our mental states cannot be exclusively reduced to swapping one set of symbols for another ( Wittgenstein , 2009 , § 146-155 ), which is all that these machines can logically do to their strings of bits .
John Searle ' s Chinese Room thought experiment uses this idea as a proof that a computer is unthinking ( I 980 ). The Chinese Room is one containing an English-speaking human operator ( Searle ) and a rulebook that gives the translations between Chinese ideograms and English text . Messages are passed into the room in Chinese , and Searle translates them using a rulebook , composes his response in English , and then retranslates his output into Chinese again using the rulebook , finally passing this message outside .
What Searle demonstrates is that it appears to an interlocutor that the inside of the room understands Chinese — for every Chineseinput , there is an appropriate and sensible Chinese output . And yet , Searle does not understand Chinese , all he is doing is translating one set of symbols into another by processing them according to rules , and then re-translating them to produce an output . This is exactly what happens inside our computers — no matter what sort of inputs and outputs they are processing , all is reduced down to applying predefined syntactic rules to symbolic operations without thought .
Nevertheless , objections arise that consider the whole system as now being functionally equivalent to thinking - we could not tell that the internals were only doing rule-following , and so we can consider the system as thinking . However , I do not agree . Wittgenstein ' s rule following considerations serve to show that this attempt at functional equivalence does not get us all the way to having the mental states crucial for thought . To continue to say that the computer as a whole system is thinking , would mean that we would somehow jumpout of the abstract ' mathematical machine ' and down into the electronics themselves . We would soon have to extend the label thinking to the power supply generating the electrons used as microprocessor ' s symbolic ' ones ', perhaps even the cables of the building and the local power station would have to be considered as part of that thinking system too , as well as the inert earth-grounded metal case that supplies the reference ' zeros ', and we would quickly find ourselves including the whole planet as part of our thinking system , which is nonsensical .
We might also question why nature did not directly evolve silicon computational substrates as the bearer of thought — in many ways the
RHYS JONES
145