on 11 August 2001
The author seeks to show that there is something about the mind that is 'not of the material world', and thus persuade us that a conscious computer or robot can never be built. But the human brain is the most complex thing in the known universe, by far, and how it works is largely unclear. One would surely need to understand it totally before properly concluding that it could never be replicated. On current scientific evidence, the brain consists entirely of physical matter organised using DNA and is the result of a process of evolution that has taken place over hundreds of millions of years. As part of his evidence, there is extensive discussion of a Turing machine, which is a primitive theoretical computer conceived before World War Two and fed using infinite lengths of paper tape. Whilst a mathematician might consider such a device interesting, its relevance to the operation of brain cells in a mouse, cat or human is far from obvious. Similarly, the Mandelbrot set - a swirling geometric pattern created by computer - is of dubious relevance in understanding what the 'I' in our consciousness really is. The author is well aware of the problem of proving the existence of free will; his suggestion that quantum effects within brain cells free us from a predetermined future has - subsequent to the publication of this book - been shown as incorrect. The belief that nothing can travel faster than light - which is included in the book and is therefore presumably part of the overall argument - has also apparently been shown, in very recent laboratory experiments to be wrong. Even if the book's central theme is incorrect, it certainly contains useful information and is valuable in helping us to understand the universe, provided that it is not taken as gospel in either the literal or metaphorical sense.
on 22 October 1998
This is a challenging book. It ranges widely through mathematics, quantum mechanics and cosmology, and it does so in depth.
But the effort is worth it. It is well set out and well-written.
If it doesn't come to any definite conclusions, well that is because the answers are still out there waiting to be discovered.
on 5 January 2010
Roger's tome was required reading for one of the courses that formed part of my master's degree. After a good introductory chapter, Dr Penrose descends into a confused mess of Turing machines, fractals, real numbers, complex numbers, philosophy, geometry, logic, neurology, quantum mechanics and cosmology in an attempt to 'prove' that it's impossible to construct a machine that can have a conscious, self-aware mind. The book bored me to distraction and I quickly realised that reading it was more effort than it was worth and I consigned it to the top bookshelf, where it stayed. I passed the course regardless.
In the intervening 15 years or so, I've made a couple of efforts to read the book again, but I'm afraid that by the time I get half way through chapter 2, my eyes have well and truly glazed over.
If you buy this expecting another Brief History of Time then you are going to be sorely disappointed. This book was written to impress other academics, who probably only pretended to have read it themselves, preferring the synopsis that one of their research students prepared for them.
on 1 November 2004
Penrose asks in Chapter I: can a computer have a mind? Having mind, he speculates, is to think or 'perhaps even to experience feelings' as a human would. He compares the mental workings of a human with the computability of machines and ultimately concludes that the breadth and power of the mind could not be replicated by a machine.
Or something along these lines. I believe that his conceptual analysis of the issues is naïve and confused and that it is so because of a fatal attempt to present his work using mainly scientific terms (lots of which are redundant and irrelevant).
How intelligent is Artificial Intelligence, therefore, and, if it is intelligent, why should we qualify that intelligence as artificial?
The central eight chapters of the book are a grand tour of some of the accepted notions of mathematical logic, complex numbers and fractals, computability, classical physics, quantum mechanics and relativity, and, finally, the physiology of the brain. We are finally returned to the main question at the end of the book supposedly informed and invigorated (or enervated!) by this immersion into a cold bath of scientific dogma having, it is implied, thereby acquired the wherewithal to solve - with the author - the problems which have been posed at the book's beginning.
I suspect that the central (and, I believe, pseudo-) question which tantalises Penrose and which has tantalised so many scientifically-minded thinkers is this: given that humans can build machines of progressive and, indeed, seemingly unlimited complexity, able to perform more and more functions hitherto only performed by humans - as, for example, the ability to play chess, how is it that at no stage of this endeavour to equal nature's complexity can we expect that the machine will become self-aware?
I believe that this is possibly a psychological not a logically robust question because, firstly, there are no limits in principle to technological advance, even to that ultimate stage when humans - armed with all of nature's secrets, including knowledge of quantum mechanics, and with unlimited power - could replicate human or human-like entities. Nature does it; it is therefore possible; why should not we be able to do the same?
Secondly, we would have the same difficulties in deciding if a machine is self-aware as we have in deciding whether animals - or, indeed, other humans are aware. This is the old notion of solipsism made fashionable for use with computers.
Minds are modes of existing, not incidental human possessions that can be placed under the microscope.
The approach of many scientists, writing books like The Emperor's New Mind which explain or popularise science or, especially, attempt to apply philosophical concepts, is to use words that are used in scientific culture (which is what all science is), such as atom, *field, *time, differential equation, and so on, as primary. That is, to take them out of a cultural context and use them as if there were no context and, indeed, as if the cultural context could be deduced from them alone and as if they may not be mixed with words from other traditions. (Interestingly, is not this what religions do?) This is why the tour of the scientific disciplines in ENM is undertaken - to burn off the semantic and procedural underpinnings of these terms and to use them in isolation to attempt to 'solve' what I believe is a collection of pseudo-problems using what are now abstract and mis-applied ideas. (These notions do - I should stress - have their place and truth in technology and operational and mathematical science.)
It is interesting that by so-doing quantum scientists are now wrestling with the difficulty - and necessity - of re-introducing the observer!
For the reasons given above I have great difficulties reading books like ENM on their own terms. One has to disregard large amounts of the surface discussion to substitute a personal interpretation of what is, in fact, happening. I finish with a quote from p.293: 'Is the presence of a conscious being necessary for a 'measurement' actually to take place? I think that only a small minority of quantum physicists would affirm such a view'