HRD's pocket guide to... machine learning
Thirza Tooes, October 26, 2018
The HRD’s pocket guide series offers an explanation of areas outside day-to-day HR that business-savvy HRDs need to have a handle on
Why do I need to know about it?
Machine learning as a concept has been around since the 1950s, but we’ve only recently reached a point where computers are powerful enough, and there is enough data, for it to be realised.
Simply put, machine learning is the ability of a machine to learn how to make predictions and recommendations by processing data and experiences. It doesn’t require any explicit human input, except to say whether a conclusion is a failure or success. It can then learn from that failure or success.
Machine learning often gets confused with artificial intelligence (AI). Although linked they’re not the same and not interchangeable. “Machine learning is an important and powerful tool that helps us design AI systems,” explains Alaister Moull, a machine learning and data science consultant at PwC. “At the highest level it’s about machines learning from previous experiences, using a variety of methods like supervised, unsupervised and reinforcement learning. We could almost imagine AI as a car and machine learning as the engine.”
What do I need to know?
Most businesses can find a use for machine learning if any of their processes involve repetitive tasks or decisions based on data.
“[Machine learning] presents an opportunity to remove simple, routine and repetitive work from people’s roles, leaving them to concentrate on adding value and making a difference to the customer experience. That makes sense from a service, commercial and employee perspective too,” says Danny Harmer, chief people officer at Metro Bank.
The bank is already using machine learning technology, and endeavours to use it in a way that neither customers nor colleagues will notice. “We use a machine learning technology that helps us provide customers with a frictionless experience. The service – which is constantly learning – uses the number calling us to match the customer to a colleague with the right skillset to assist them,” Harmer explains. “Customers obviously aren’t aware of the complex system working in the background and instead – along with colleagues – have a seamless experience.”
However, this won’t necessarily be an easy transition. Getting machine learning off the ground is a big technical undertaking, points out Zarek Rahman, an independent solicitor who specialises in advising technology companies and startups.
“The business needs to have a large set of data for the computer to analyse before it can make such predictions, and that data needs to be cleaned up and put into a format that the computer can understand. This can be a massively time-consuming and expensive process,” he says.
Where can HR add value?
It might also not be an easy transition for employees. People may resent work being ‘taken’ from them by machines, particularly if they’re made redundant rather than redeployed.
Ensuring that employees, and by extension customers/service users, can work alongside machine learning technology will be key. This will likely involve upskilling, retraining, redistributing and reassuring the existing workforce.
“We need to provide our people with the skills required to work alongside and complement the work of machines,” states Gillian Morris, global head of talent management at Jaguar Land Rover. “While old types of jobs might disappear, new ones will be created. We will need to harness the multifaceted, adaptable and creative intelligence of humans. As well as providing opportunities to learn new skills, HR will need to offer support to their people in adapting to this change.”
While there have been huge advancements in recent decades, and doubtless we’ll see more in the years to come, it’s not time for humans to hang up their hats just yet.
“Right now AI is a tool that can be used to speed up existing and archaic processes. While this might lead to the loss of traditional roles it also opens up new ones, primarily under the guise of ‘data scientist’,” points out Rahman. “Within my own field machine learning is being used to analyse historic legal judgements to predict the outcome of a case; something that might traditionally be done by a trainee lawyer. This does not mean that all those training to become lawyers will suddenly find there are no jobs, but [they] need to understand how to teach machines and essentially take on the skills of a data scientist to do so.”