Artificial Intelligence

WHAT IS ARTIFICIAL INTELLIGENCE

​

One of the most important aspects of AI is that it is a multi-use technology. Like electricity it can be appliedin lots of different ways, to lots of different scenarios.

 

There is no single, universally accepted definition for Artificial Intelligence, but the Oxford English Dictionary defines AI as “the capacity of computers, or other machines, to exhibit intelligent behaviour”.

 

This means AI systems appear to think, learn and act like humans and in some cases exceed the capabilities of humans.

 

AI systems can analyse vast amounts of data, solve complex problems, make decisions and perform creative tasks.

​

Some AI technologies have been around for more than 50 years, but advances in computing power, the availability of enormous quantities of data and new developments in software algorithms have led to major AI breakthroughs in recent years.

 

It is these three components of advanced algorithms, data and computing power, that explain how machines can exhibit intelligent behaviour and why AI has suddenly exploded into our everyday lives.

​

ALGORITHMS

 

THE BRAINS OF AI ‘Traditional programming’ involves encoding human knowledge and experience into a set of precise rules that a computer can follow, step-by-step, which make the computer appear to respond intelligently. These rules, called algorithms, tell computers how to perform tasks and in traditional programming are often expressed in an ‘IF-THEN-ELSE’ format, which resembles a decision tree.

 

For example to create a ‘digital doctor’, an algorithm might look like:

 

IF the patient has a fever,

THEN prescribe Drug

ELSE send the patient home

 

The intelligence in traditional computer systems comes directly from human knowledge and expertise being recorded into a format that a computer can process. Humans can read this ‘computer code’ and understand how these digital systems make specific decisions and can therefore identify mistakes and improve the system. For example, in the ‘digital doctor’ scenario, new clauses could be added to deal with special cases, or to reflect new medical knowledge.

 

For example:

​

IF the patient has a fever AND is allergic to Drug X

THEN prescribe Drug Y

ELSE prescribe Drug X

​

This example now starts to illustrate a key limitation of traditionally programmed systems.

To develop a useful and reliable ‘digital doctor’, a huge number of rules and exceptions would be required that the system would very quickly become very large and very complicated.

 

Thousands of IF-THEN-ELSE rules would still be unlikely to capture all of a real doctor’s expertise and experience gained over time. Therefore traditional programming of ‘intelligent systems’ is best suited to constrained environments which do not change much over time and where the rules can be strictly defined.

​

A good example of this could be an intelligent ‘tax calculator’ which when presented with an individual’s financial income and other relevant circumstances could efficiently review a series of ‘IF-THEN-ELSE’ rules to calculate their tax liability