• Contribute
• What's New
• Submit News

	6.5
	Author: David Harel Excerpt: Computers are incredible—one of the most important inventions of the last hundred years...As we continue to enjoy the benefits of these amazing machines, it is time that we took on board some bad news. .: Buy at Amazon.com :. .: Buy at Amazon.co.uk :.

As an artificial intelligence enthusiast, I am always reading about what computers will be capable of doing. From running our phone systems, drawing pictures, computer games, writing poetry and an array of others goals of AI. I thought, though, that it was time for me to see the other side of the picture, what computers cannot do.

Harel's new book Computers Ltd. seemed like an excellent place to start. Harel's book basically starts by laying down the age-old problem that Turing set out to solve - is everything computable? As Turing (and Church) found out, there are many things that are simply not computable. Harel has a field day on this fact, bringing up numerous examples and showing just how useless a computer would be in such a situtation.

The examples are all interesting by themselves, but Harel brings so many up in detail and goes into the same points over and over again, the book can often really slow down. Thing is, he then goes on to show how some things are computable, but just very inefficient (either due to time or memory) and lists numerous examples which are also a little redundant, in my opinion. Again, all the examples are interesting by themselves and could make interesting reading for a textbook style book that doesn't require everything to be read, but in Computers Ltd. you find yourself saying "Yeah, yeah. I get the POINT!"

Luckily, the book is saved by two rather good chapters toward the end of the book. There is a good chapter that covers the new computing paradigms: parallel, quantum and molecular. Some are covered in more depth than others, but I do think that Harel didn't quite show some of the amazing advantages that they can have over the current serial computers. Sure, he points out that quantum computing can render current crytography obsolete, and that molecular and parallel computing can utilize cooperative computing to create efficient algorithms. There is still a lot more exciting information about those fields I felt he didn't divulge.

There was also an excellent chapter on cryptography, which beautifully explains RSA encryption, how it utilizes the weaknesses of modern-day computers. It also covers digital signatures and a few other small points. This was definitely the highlight of the book for me, since it was all new information and it presented the faults of current computing in a palatable form.

One of the final chapters covered Artificial Intelligence - oh dear. It was really made up of ludicrious hypothetical situations that made the computer look stupid. The chapter also covers the Turing Test and how hard natural language is, but both of these have nothing to do with non-computablility, they just seemed to be in there to further show how stupid computers can be.

Overall, this book did have some true gems of information, but it was not presented in a decent form — it was just too biased. I will be the first to admit that computers can be infuriatingly dumb at times, but do not see how the fact that a computer cannot reasonably factor a 200-digit number into primes can used as a reason against their usefulness.

Not really for people interested in Artificial Intelligence, but people who feel comfortable knowing what computers will never do will love this one.