Artificial Intelligence: an Introduction

by Sarnaz Hossain

(image: Franck V)

Artificial intelligence is a wide branch of computer science concerned with building smart machines capable of performing tasks that typically require human intelligence. AI is already hugely prominent powering things that we take for granted on a daily basis. These include things such as search engines, digital chat boxes. The main basic idea of AI is to be able to analyse large amounts of data and therefore learn to complete a certain task. This is termed as machine learning.

However interesting and advanced AI is today, it is important to remember that it is not a cutting edge new discovery. Alan Turing in the twentieth century developed a test called the Turing test which examined whether a machine has the ability to exhibit intelligent behaviour equivalent to a human. Turing predicted that by the twenty first century AI would have advanced so much that they should be about to pass this test with flying colours and have the ability to mimic human intelligence however only 2 well known programs claim to have passed this test. This shows that we still have a long way to go in AI development. At the moment we have reached the point of Artificial Narrow Intelligence which includes AI such as Alexa and recommendation algorithms. These types of AI can solve simple problems but don't have “consciousness” or ability to adapt to different tasks. The next level is called Artificial General Intelligence which is AI that have the ability to overcome the Turing Test and be able to pass as having human enough intelligence. Even though we haven’t got to this stage just yet with the advancements of self driving cars and reinforcement learning we are growing closer and closer to this reality and by the next few years we are expected to see a huge advancement. The final level of AI is Artificial Super Intelligence and so far this is a very abstract idea that after a while AI intelligence will actually surpass human intelligence. An idea that has been delved into by movies and dramas, possibly a little overboard, connoting ideas of take over and devastation. 

A very common question when talking about machine learning and artificial intelligence is how a computer can actually learn. To set up a very simple AI machine there are 7 key steps. Firstly it is necessary to define the objective of this machine. This includes the target variable for example, when creating a weather predicting machine, the target variable would be whether it would rain tomorrow Y/N. Then it is necessary to collect the data needed. This can be manually inputted data or more commonly online resources. After that the next step would be to prepare the data this means clearing any redundancies in the data and removing incorrect data. In addition while this is happening it is important to analyse the data understanding patterns and relevant information as a computer is not very good at understanding what data is relevant and what has no effect. Going back to the example for predicting weather, the relevant information would be things such as humidity, temperature whereas information such as the day of the week is not relevant to whether or not there is going to be rain. Next you need to build your model. This will be done using a Machine learning algorithm for example, linear regression or logistic regression.

Finally after the model is created it is necessary to evaluate the efficiency and accuracy of the machine and also the final predictions.