Artificial intelligence (AI) is a form of computer science that is built to have machines think and respond like humans, understanding natural language and reasoning with data in a way similar to the human brain.
It refers to programming computer technology, known as machine learning, to complement the human mind when making decisions, making make our jobs easier and minimising repetitive tasks.
For some businesses, AI still appears to be in its infancy. However, the technology definitely has the potential to transform many organisations as well as the technology industry as a whole. AI has been adopted to automate a number of roles in businesses.
AI emerged from the ‘Summer Research Project on Artificial Intelligence’ workshop at Dartmouth College in 1956 by a number of researchers. John McCarthy and Marvin Minsky first came up with the term and are recognised as the founders of AI research.
Machine learning was coined by Arthur Samuel, an American pioneer of computer gaming and AI, while working at IBM in 1959. Years later, the term deep learning was coined by Rina Dechter – a computer science professor – in 1986.
Uses of AI
Believe it or not, many of us use AI on a daily basis without even knowing. AI is integrated into a lot of smartphones and smart devices.
Apple’s Siri and Microsoft’s Cortana are built using voice processing to act as personal assistants and this is through the understanding of natural language.
Another example is Amazon’s Alexa, an intelligent voice assistant built with neural networks for natural language processing first developed in 2014.
AI has evolved over the years to be used in industries ranging from healthcare to retail – for instance, Nvidia’s 2017 partnership with Nuance to bring AI into radiology reporting for diagnostic imaging.
Machine learning vs Deep learning
Two popular forms of AI are machine learning and deep learning, both refer to applied AI which has had the biggest impact on businesses so far in terms of adoption.
According to Salesforce, machine learning is the 'core driver of AI'. It is the concept of having computers learn from data.
Examples include using algorithms to spot details that may be hard for humans to identify; this is widely deployed in finance and healthcare. In finance, machine learning is used to predict bad loans and credit scores.
However, machine learning is mainly reliant on human prompting and over the years more and more sectors require machines that could perform duties in ways similar to the human brain – which brings the rise of deep learning.
Deep learning is a kind of AI that uses complex algorithms to perform tasks with minimal or no human prompting.
The term refers to the development of neural networks that resemble the human brain. This involves training a computational model to understand natural language.
Supervised and unsupervised are two machine learning methods that describe how algorithms are trained on a data set and expected to learn from it. Supervised tends to be the most commonly adopted.
Using supervised learning, the algorithm output is already made clear during the training stage; it maps out the input to output.
Unsupervised learning, on the other hand, is a more complex process. It does not include a training data set and the outcome is not known. It can gather unstructured data from various sources.
A widely adopted example of deep learning is image analysis, whereby a computer is taught to classify images by analysing a range of other images and data points – for example, Google Photos uses deep learning to power facial recognition in photos.
A neural network is a computer system that is built to resemble the human brain and nervous system. The human brain is often modeled around a network of neurons, known as biological neural networks.
Also described as artificial neural networks (ANNs), these are one of the main tools used in AI and machine learning to help support the machines learning the way humans do.
Neural networks include input and output layers, as well as a hidden layer of units that transforms the input into output. They are good for finding patterns, particularly in unsupervised learning, in complex processes.
Advantages of AI
As mentioned above, AI provides a huge amount of potential for business, finance and healthcare sectors, especially with the assistance of repetitive or mundane tasks.
“From backend IT service desk tasks to interactive, customer-facing roles, cognitive AI agents are matching real human capabilities, making people’s lives easier and improving overall productivity,” Martin Linstrom, managing director UK&I at IPsoft tells Techworld.
“By freeing up employee time to focus on more innovative tasks that require a human touch, AI is creating a modern workforce that is better equipped to tap into their creative potential,” Linstrom says.
AI is also a great route for businesses to automate, make decisions faster and avoid human error.
Disadvantages of AI
AI has raised fears among people who view it as a threat – particularly in the form of job losses.
A 2017 report from PWC suggested up to 30 percent of UK jobs are at high risk of automation by 2030.
“In spite of tremendous recent advances, AI is in its infancy; we are still blind to many of the ethical dilemmas that this technology will impose on our lives,” says Jonathan Ebsworth, speaking with Techworld.
“Existing laws and codes of ethics were not designed with an AI-enabled world in mind, and while it’s clear that we need guidelines and codes of practice to ensure that humans are not harmed by new technology, what we need most of all is for companies who are working with AI to take full, informed ethical responsibility for the solutions they deploy.”