Главная arrow Языки arrow Английский язык arrow


It is quite possible to set out an approximate scale of intelligence: most people are more intelligent than most chimpanzees, a word processor is a more intelli­gent machine than a typewriter, et' Nevertheless there is no scientific definition of intelligence. Intelligence is related to the ability lo recognise patterns, draw reasoned conclusions, analyse complex systems into simple elements and resohe contradictions, yet it is more than all of these. Intelligence is at a higher level than information and knowledge, but below the level of wisdom. It contains an indefinable 'spark' uhich enables new insights to be gamed, new theories to be formulated and new knowledge to be established.

Intelligence can also be examined from the point of view of language. Infor­mation can easily be represented as words, numbers or some other symbols. Knowledge is geneially expressed in a language or as mathematics Intelligence is at the upper limit of language instances of patterns or deductive reasoning can be written doun. and certain general principles can be stated However, the creative 'spark' of intelligence is almost impossible to evprcss in language

The only widely accepted definition of artificial intelligence is based on a test devised by Alan Turing in 1950.

Suppose there are two Identical terminals in a room, one connected to a computer, tnd the other operate remotely bj a person. If someone using the tw> terminals is unable to decide which is connected lo the computer, and «hich is opuuu-d t>> на- >.4i-*,,,, .1 v. ' ;? ; '?' "i *•«• rrH'U-i *iih intellieence.

1 he definition of artificial intelligence which follows from this lest is:

Nd computer system has come anywhere near to passing the Turing test in general terms. Nevertheless, progress has, been made in a number of specific fields. It would take a very good chess player in the 1980s to be able lo tell whether he or she were playing agzinst a computer or a human opponent. Most car drivers are unaware which pans of their cars have been assembled by robots, and which by manual workers.

Conventional data processing is bised on information; artificial intelligence is based on knowledge. A central problem for artificial intelligence is an adequate representation of knowledge on a computer. On the one hand, the representation must be 'rich' enough to be of practical use. On the other hand, it must be simple enough for processing by i computer.

A technique of knowledge representation which is widely used is semantic networks. As shown in Figure '.I. i semantic network shows a set of relation­ships between objects. It is a flexib!e method of representation, allowing new objects and new relationships '.o t« added to a knowledge base. Accordingly, semantic networks are often u«d in computer systems which have some form of learning capacity.

Gam* Playing Programs

Much of the progress in artificial intelligence has come through work on game playing programs. Games such as chess have the advantage of being simple enough to represent on a corrputer. while requiring a high level of intelligence on the part of the player. A r. jmb<r of successful strategies for playing games have been worked out. They are all based on searching a large number of possible moves and counter-moves, and selecting the best one to make. In some games, such as noughts and crossev it is possible to search right through to the end of the game for each possible r ;xt move. In other games, notably chess, this is not possible, as the number of moves is too large even for the most powerful computer. The best chess programs achieve the right balance between the breadth of the search (the number of possible moves investigated), the depth of ». -.».. —л,, , г глч«.*м>«п| moves investigated for each possibility)

and trie way vi дьмьмнф ни. •«•«.. ,

Many of the methods used for game playing programs are being transferred to o'Vt fields cf artificial intelligence.

Reasoning programs have been u^d to solve the kind of pattern recognition problems found in intelligence tot*, and to solve problems in formal logic. An example of programs of this *>ti is the use of a computer to assist in the proof of the Four Colour Theorem. It has been known for centuries that no more thanfour colours are needed to colour in any map, so thai no two adjacent' zones have the same colours. See Figure 7.3. This theorem was finally proved with the aid of a computer program in 1976.

Computers cannot interpret

continuous passages in a natural language. Nevertheless, computers can cope with individual words and phrases, and longer passages of natural language in specific topics. A major topic of artificial intelligence research has been the recognition of natural language by computer. There are two aspects of this work, syntax and semantics.

12 Natural languages are composed of structures such as sentences, which areconstructed according to rules of syntax. For example, the sentence:

The boy stood on the burning deck can be analysed (or parsed) as

<subject> <verb> <object> where < subject > (the boy') can be further parsed as

<article> <noun> etc.

The problem with syntax analysis is that the rules for sentence construction are very complex, there are many exceptions, and the rules are gradually modified as languages evolve.

  1. In order to understand a passage in a natural language, the semantics or
    meaning of the piece must he studied. This depends on the context and what has
    been said before, as well as the meanings of individual words. Semantics is very
    difficult. In some cases an alternative interpretation of a single word can alter
    the meaning of a whole passage

  2. Computer programs have been devised which will cope with the syntax and
    semantics of complete sentences, but only within limited contexts. Even for
    these restricted situations, the programs are very complex. However, if current
    research into fifth generation computers is successful, systems with a much more
    powerful natural language capability will be available during the 1990s.

Если Вы заметили ошибку в тексте выделите слово и нажмите Shift + Enter
Культура. Искусство
Политология. Социология
Религия. Логика. Этика. Философия
Сельское хозяйство. Биология. Ветеринария.
Финансы. Экономика