13 March 1998
A fictional encounter between leading academics at the dawn of the computer
age provides an ideal environment in which to air many of the issues
concerning the field of Artifical Intelligence. This work is labeled as part
of the genre of `scientific fiction'. It attempts to use a fictional setting
to convey issues relating to human cognition and the possibility of machines
being able to reproduce or mimick this phenomena within the context of the
bounds of machine capability at the period in which the story is told.
By creating this fictional setting, Casti is able to seamlessly intergrate many of the issues relating to the subject and consider the perspectives that some of the greatest minds of the period might have held. Though much is conjecture, Casti is able to extrapolate from our knowledge of the works of In order to create an encounter and discussion, Casti places the responsibility on C.P. Snow a physicist, novelist and acting advisor to the government on science policy, to perform in 1949 an informal investigation into the feasability of artificial intelligence (This term was actually first recorded to be used at the Darmouth College meeting of 1956 by John McCarthy).
Snow chooses to invite four prominent academics: Alan Turing, one of the founding researchers of Computer science who was later to be considered a founding farther of artificial Intelligence. Ludwig Wittgenstein who was a leading philosopher in the fields of language use and other minds. J.B.Haldane a biologist who contributed to explanations of evolutionary theory through classical genetics. E. Schrodinger a nobel laureate physicist but also at this stage a pioneer in the science of Biochemistry. Snow chooses to host the meeting over a dinner party. By placing the context of the discussion in the informal situation of a dinner party, Casti feels able to push the points of view of the participants further than might be expected in a more formal debating chamber (though with the quantity of wine that they quaff, it suprises me that they are able to remain quite so coherent whilst finishing their disscussion over cognac and cigars). The structure of the book is broken down into chapters based around each course of a particularly sumptuous meal ( for postwar rationed Britain ).
Over a sherry, the characters introduce themselves with an account of their work. It is here that Snow also reveals the purpose of the evening. With the starter, Turing introduces the notion of computation which is still relatively unknown outside the field of pure mathematics. The notion of a Turing machine is explained and the idea is put forward about what such machines are able to compute. From here, Turing is able to postulate the idea that mental states are analagous to solutions from Turing machines .This arguement particularly relies on Turings experience of code breaking during the war and chess playing for which he had given considerable thought to.
With an explanation of the mechanism of how such an Artificial Intelligence might be realised, the next course (fish) centred around the idea of what exactly intelligence might be for any kind of assessment to be made about the machines capabilities. It is here that the famous `Turing test' is proposed as a means to measure the intelligence of a machine. Issues relating to this such as the idea of whether it is really fair to expect a machine to behave exactly like a human were aired but a more concerning issue arrose which was discussed over the next course.
With the meat came the issue put forward by Wittgenstein that the manipulation of symbols alone did not provide the machine with any real semantic knowledge. This issue was further explored with an arguement based aroud John Searl's Chinese Room. As with the Chinese Room debate much time was spent exploring the possible fallacies of this approach, leading on the the issue of language.
During the next course, Language of Thought issues were raised and the possibility of language acquisition was further explored. The debate hung on ideas that were infact developed some years later by Chomsky on how language acquisition devices might come to exist in humans. A further point put forward here was the idea of language being a shared experience between humans and how a machine would have quite a different perspective and thus no tangible hold on human concepts.
The dessert moved the question further away from the central theme of intelligence to consider what life itself was and whether computers would be able to qualify for this category. Appeals were made to the idea that with Cartesianism rejected and with a materialist perspective there was no reason to imagine that the many characteristics enjoyed by living organisms could not be denied to a computer system. >From this followed further questions as to the possible cultures that might arise from such Artificial Agents.
The story concluded with the participants going their seperate ways, Turing and Wittgenstein still opposed to each others views and unchanged by the discussion but with all three of the others if not converted to the idea that machine intelligence might one day become a possibility then certainly aware that the debate it could cause would open up many more questions about human identity and cognition.
The final chapter gives a brief overview of the history of A.I. up to the current time highlighting the proper sources of many of the issues raised. Having been provided with a taste for the subject this is an excellent source of reference to correct any missunderstandings and to pursue any particular arguement in a more rigorous fashion.
To consider what sort of readership this book might appeal to, the introduction demonstrates clearly one of the strenghts of this form of exposition. As the concepts at the time were quite new, everything has to be explained to all of the other characters familiar only with their own particular field. This allows the author the opportunity to lay out all of the material neccessary for such a debate in a natural and interesting style. This book would provide a very readable overview of many of the issues at the heart of the A.I. debate for someone entering the subject. However, caution must be exercised in placing too much trust in the text as many issues are not clearly given their true source such as Searle's Chinese Room arguement.