Origo is here to support you during this new normal. Get the latest updates in our blog
Back in 1949, an American mathematician named Claude Shannon published a paper entitled ‘Programming a Computer for Playing Chess’. It was a landmark paper and, if you’re really interested in algorithms, you can find it on the internet, but you don’t need to read it to understand its significance. The point about Shannon’s paper is that it shows that there is nothing new in the concept of teaching computers to learn for themselves – not only to execute complex calculations but to learn and improve as they do so.
Machine learning has been around for decades. In 1997 an IBM computer named Deep Blue vindicated Shannon’s proposition by beating World Chess Champion Garry Kasparov at his own game. Because of the cost of data storage and slow processing speeds, however, machine learning was slow to make much impression beyond the fields of sci-fi and chess.
Times have changed. Large quantities of data are readily available to businesses of all sizes and you don’t need a computer the size of a house to process it. So how could machine learning benefit your business?
Today we are surrounded by applications of machine learning. Here are some examples:
Ever feel like your computer knows you better than you know yourself? Those uncanny shopping suggestions that pop up from Amazon are the fruits of a machine learning about your tastes and buying habits. Every time you browse or buy, the computer takes note of what you’re looking at, where you go next, what actions you take etc and formulates an increasingly accurate understanding of your behaviour. A human could shadow you for a while and come to similar conclusions, but the Amazon algorithms are doing this in seconds for millions of transactions every day and the learning curve is exponential.
The future of transport is unwritten but it’s fair to say that the way we get around today is stupidly inefficient compared to what the future holds. Machine learning is helping to manage traffic, making rapid calculations and applying fresh learnings from constantly updating data sources to apply controls that keep the lines flowing. Logistics companies are using machine learning to calculate the optimum transit routes for their customers’ goods at any given time. And self-drive technology, which uses machine learning to enable a vehicle to plan its own movement according to its position and what’s around it, is being applied in aviation, shipping and autonomous cars like the Google Waymo.
Machine learning is being applied across the healthcare spectrum, from the management of patient flows in hospitals to the assessment of patients’ conditions and more accurate diagnoses. Doctors make their diagnoses by assessing the symptoms, gathering historical information and observing patient responses to certain tests. But they are still reliant on their scope of knowledge, which can never be comprehensive. A computer trained with the same information can scan a far greater range of possible causes and provide doctors with suggestions that they might never have considered.
The banking industry is using machine learning to identify patterns in customer behaviour which can be used to provide an early warning against fraudulent transactions. These patterns become increasingly sophisticated each time a transaction is challenged and the machine learns from the outcome. As a result, the machine can identify when an anomaly fits with a particular customer’s behaviour and thus reduce the number of false positives. At the investment end of banking, machine learning is used to calculate risk and identify opportunities.
Similar to the way Amazon presents you with shopping suggestions, Netflix, Pandora and other entertainment platforms have been using machine learning for several years to gauge customers’ viewing and listening tastes and offer suggestions accordingly. The devices we use to access these platforms, such as Amazon Echo, are also examples of machine learning. Every time you ask a question and follow it up with a decision, the machine learns something new about your tastes and behaviour and becomes better at catering for your needs.
It can be daunting trying to understand the potential of machine learning and how it could benefit your business. Like Shannon’s paper, there is a lot of information online written by programming wizards with a language all of their own, who seem more keen to discuss the complexities of algorithms than to reveal the simple steps you need to take to make machine learning work for you.
The best advice comes from, perhaps, a surprising source. Danny Lange, who led the machine learning team at Amazon and now heads up machine learning at Uber, is very encouraging for businesses of all sizes to start using machine learning. He offers five rules:
Start now! There is no reason to wait. Once you get started, you’ll realise that machine learning is not some futuristic fantasy that only the giants like Amazon, Uber and Netflix can exploit. Any business can use it to improve.
Start with something simple like predicting customer churn. Many companies lose customers on a regular basis, regard it as a fact of life and focus on winning more business than they lose. But with intelligent customer attrition analysis, fed by historical data, you can predict which customers are likely to be thinking of leaving at any given time and take steps to keep them on board.
It costs a lot more to win a new customer than it does to retain one, so this is one simple machine learning model that could transform your business performance almost overnight. Machine learning is out there – why wait?