New model argument

十二月 15, 1995

So is Roger Penrose (THES, December 8) an opponent of computer science, pathetically fighting for the superiority of physics over computing? I wonder what has caused this ludicrous bias? It seems to me he does not understand this subject.

For example, he believes he can predict the future of computer science, making statements such as "human beings will always be cleverer than computers". This statement cannot be proved with any present science, therefore it can only be bigotry. He betrays the simplicity of his arguments with his attack on the field of neural networks. No sensible computer scientist would stake the future of artificial intelligence solely on building bigger neural networks. What about new methods such as analogical processing, fuzzy natural language system, genetic algorithms and methods yet to be discovered? Does he not realise that neural nets are just part of our development of a better science?

He tries to deny his need for the specialness of man by admitting that some science other than computing might model the human mind. How can this not draw on and hence be part of computing? No matter what methods are used to explain consciousness it will most likely have to be modelled by some form of computer. Computers can model any form of organised, chaotic or random behaviour, where a model exists. The future of computing will not be in the current architecture of neural networks but systems developed from the knowledge gained by each new development in computing.

He states that people "have not faced up to the implications of mathematical logical thought". Where are his mathematical or logical proofs? Computers are modelling machines that are used in all sciences to model any behaviour. Those who use this kind of pseudo-science to discredit the abilities of computers are building and admiring the emperor's new philosophy.

Martin Lefley Senior lecturer, applied psychology, Bournemouth University

请先注册再继续

为何要注册?

  • 注册是免费的,而且十分便捷
  • 注册成功后,您每月可免费阅读3篇文章
  • 订阅我们的邮件
注册
Please 登录 or 注册 to read this article.