It’s one thing to build a machine capable of intelligent behaviour (AI), it’s quite another to get it to act without being explicitly programmed. It’s this latter science of Machine Learning that’s sparking so much interest amongst my old banking colleagues in San Jose and San Francisco. I returned recently at the invitation of Copenhagen-based Danske Bank, to meet with old colleagues, new friends and attend a series of events given by leaders of the tech community over there. The epicentre of all discussion and debate was Machine’s Learning application of data. The power of the machine to learn, detect and make high-value predictions about consumer behaviour based on real-time modelling.

Data unites the hot topics. Wearables, the Internet of Things, continuous authentication, the sharing economy, 3D printing, and of course personalised audio and video content. Those with money to invest in today’s start-ups are looking for companies turning data into clever new services. As you can imagine, the ability to predict future demand is gold dust in the right entrepreneur’s hands.

Here are some of the inspiring examples I heard. I believe, if you’re a service provider, they make a solid case for Machine Learning to move front and centre of your strategic planning agenda.

The power to predict

Machine Learning algorithms have already woven their way into different aspects of our lives, whether it’s getting an accurate eta for an Uber taxi, or Netflix recommending the films to watch next, or companies like Black Sage trying to protect us from errant drones breaching security from above. On the surface, these entrepreneurial start-ups are simply providing a useful service. However, what makes them remarkable is that underneath they are also collecting masses of data to expand their insights.

Square in the US provides SMEs and small entrepreneurs with a mobile card reader they can take anywhere. It’s a convenient service, but the real beauty is that over time the firm can see when the small businesses might need a working capital loan or line of credit, and can suggest it to them before they even realise they need one themselves.

Insurance giant Woolworths has collected data revealing customers who drink milk and eat red meat are a lower car insurance risk than those who eat pasta and rice, drink spirits and buy petrol at night. Their machine learning algorithms are building models which can predict risk based on patterns and correlations starting to appear.

Zest Finance, the US parent company of online lending company Basix, suggest that all data is credit data. They have discovered that people who fill in online name fields in CAPITAL LETTERS represent a much higher risk. Careless use of Caps Lock probably won’t mean you’re declined for a loan, but it might trigger a second layer check. As a financier, I find it fascinating that your digital writing style can provide clues to your risk profile.

Clearly, Machine Learning brings you the ability to make powerful consumer predictions. How could it radically improve or evolve the services you provide? And how do you ask for that data without provoking a customer backlash as Spotify did when it overstepped the mark?

In May, the Economist declared data to be the fuel of the future. “Data are to this century what oil was to the last one: a driver of growth and change.” They label data the currency of a new economy. “Facebook and Google initially used the data they collected from users to target advertising better. But in recent years have discovered that data can be turned into any number of AI or cognitive services, some of which will generate new sources of revenue. All of which can be sold to other firms to use in their products.”

How are you collecting data? Is it the right data? Are you nurturing it to provide the business insight you need?

Dealing with unstructured data

Data flows from our everyday actions like a constant digital vapour trail. Behaviour past and present, geo-location, social information about groups and friends, likes and so on… it’s never-ending. So how do you figure out where to start with Machine Learning? The answer it seems is to start small and work up as with every other human learning endeavour.

Dr Danny Lange, VP of AI and Machine Learning at Unity Technologies gave an example using a robotic snake and a block of wood. The snake is told to learn how to get over the piece of wood. If the piece of wood is too high to start with, the snake struggles to learn. If the wall is set low to begin with the snake learns quickly and then applies this learning as the wall increases in size. With Machine Learning the key is to gain a basic understanding of how people buy first, gradually increasing the layers of complexity over time.

Automated decisions

It’s also time to get comfortable with computers making decisions too. I only have to think about my recent brush with a travel insurance claims process to think this can only be a good thing. Rather than working from a start point that I am a deviant trying to pull a fast one, a computer can be programmed to look at my claims history and check event data linked to the flight I claim was delayed. The machine can then make a calculated decision in nanoseconds and issue a request to pay the claim. An altogether more life-affirming customer experience delivered by a machine.

Get to grips with the ethics early

How we use data to make predictions has obvious ethical challenges. To give you an example, a young Australian entrepreneur has developed an algorithm capable of detecting early onset Alzheimer’s by tracking writing style on social media (his mother developed the illness in early in life). The application can detect early changes long before the symptoms would be clear to medical professionals. This data in the hands of the NHS would be amazing and treatment could begin earlier. In other countries, it could be more problematic if, for example, health insurance companies were to use the information to push up someone’s premiums. Find out where the debate is happening in your sector and get involved early.

Make a plan

I know from experience that the tech topics capable of getting bankers talking are usually worth exploring. Whether that’s in Silicon Valley or our fintech community in the UK. Machine Learning is here and it’s already working hard for some of the world’s biggest and fastest growing start-ups. I predict you’ll think of a brilliant way to use it.