Artificial Intelligence Essay

In this essay I will attempt to describe briefly what artificial intelligence is, its drawbacks, and to what level computers have advanced in order to be able to achieve the goals of AI. In doing so, I will go on and discuss the following points: • Computers can only do what they are told by a programmer • Computers cannot think • Intelligence cannot be understood.

There are many definitions of artificial intelligence (AI), but they all conclude in some form to say that it is the study of mental capabilities through the use of computer models. It can also be said that AI allows computers to carry out tasks that have been considered to require intelligence and are usually carried out by experts. The main advantages of AI are therefore that it can aid experts in decision making and comment on any statements made, helps answer the questions of trainees requiring the knowledge of experts in their field, and obviously, learn from its mistakes to acquire new facts. When defining artificial intelligence, it is useful to understand fully what both words mean. Something artificial is said to be unreal, unnatural and inorganic meaning it being synthetic. Intelligence is the skill of understanding and the ability to perceive and conceive meaning Therefore artificial intelligence is a way of providing man made computers with the ability to learn, perceive and think for themselves. This is just the theory, because in practice, many problems arise. Firstly, computers are pre-programmed using a set of simple rules and instructions. This gives them hardly any ability to think for themselves, requiring unlimited user input and causing the need for very long and tedious programs or algorithms to be written up to provide the computer with simple logic required to understand basic facts essential in any AI application. Programs would have to create a close restricted world in which the computer would store what it has learnt allowing for mistakes that would be corrected. Secondly, computers are only intelligent in their specific field and it would take huge processing power and many man hours of programming to develop a computer intelligent in a number of fields with many problems being encountered on the way.

‘Computers can do only what they are told by a programmer.’ This is an interesting statement that portrays computers as very simple machines with hardly any intellect. In most cases it is true, as computers are very dumb machines requiring a maximum level of user interaction to do something useful. After all, computers run programs that are put together by programmers to do specific tasks. It is also true that programs obey a large number of instructions set out by the computer or architecture to remain error free. But when designing an AI system, the main aim is for the computer to be able to think by quickly reacting to the user’s input with minimal interaction required from the user to get a result. AI therefore tries to make computers intelligent so they can interpreting a situation in the smallest amount of time possible before coming to a conclusion.

‘Computers cannot think’. There are many views over the issue of a computer’s knowledge and if it can really think. To put the statement into context, the ability to think is the ability to consider, judge, or believe. These are natural abilities possessed by all humans, and as a computer does not have a biological brain, it is perceived not to have the ability of thought. This is true to some extent, and is helped by the results of the Turing test. The Turing test was devised in 1950 by the British mathematician Alan Turing as a benchmark test to show how intelligent a computer really was. But even today, as we near the 21st century, computers still cannot pass the Turing test to a convincible degree.
Another factor which computers lack is basic human logic and the difficulty of understanding simple tasks. This cannot be taught to a computer as children gain this automatically through experiences, repetition and education. However Natural Language Processing(NLP) is helping computers by injecting basic logic into them through the use of AI applications and is also making them more user friendly. On the other hand, a computer has a main processor (sometimes called the brains of the machine) which can execute a few million instructions a second. This is too advanced, even for the human brain with regard to arithmetic, but is required for the computer to be able to carry out simple instructions. Artificial intelligence may fool humans into believing that computers can really think, when all they are doing is using their immense processing power to follow a set of algorithms and come to a simple conclusion based on the information they have.

‘Intelligence cannot be understood’. This is true in some sense as it is a natural ability possessed by all humans and cannot be put fully into context. Intelligence can be defined as the capacity for understanding, but it is extremely difficult to measure intelligence in relation to human beings due to the complexity of our brains. With computers, it is considerably easier to measure the intelligence they possess. One way is by using the Turin test which has been fairly successful in displaying the downfall of complex AI systems, although the test is nearly 50 years old and new more complex tests that have been developed may be more useful.

To conclude, I believe that artificially intelligent computers will never replace experts in their particular field, partly because they are the ones who provide the computers with all their knowledge in the first place. The main goal of AI, that is recreating the human brain inside a computer, is practically unachievable at the current state of technology. Artificial Intelligence is also still in it infancy, but observing the speed at which technology has progressed in recent years, with regard to processing power and the new capabilities of modern super-computers, it won’t be long before AI becomes a reality.