The Blog

The Three As Of FinTech. AI's Evolutionary Triple Threat

Financial services, for example, has been flipped on its head thanks to AI. No more than a decade ago it was unthinkable that a customer could have a loan or credit card application underwritten online in milliseconds without an underwriter in sight.

Artificial Intelligence (AI) is here. Not in the same way that George Lucas saw it, but it is ingrained in the fabric of our daily lives. Financial services, for example, has been flipped on its head thanks to AI. No more than a decade ago it was unthinkable that a customer could have a loan or credit card application underwritten online in milliseconds without an underwriter in sight.

AI's new looming position in the public consciousness is exciting and concerning in equal measure. More than half of British consumers are worried about the impact robots and AI will have on our lives, with many believing they will destroy humanity as we know it. The Government's latest Science and technology report took the first step in recognising AI's potential impact in the future of Britain, but until we see robots taking over the streets of London there won't be any regulation put in place to control it. And rightly so.

Below are three of the most material themes that have sat at the intersection of FinTech and AI and will continue to define it into the future - Arithmetic, Abstraction and Anticipation.


By 2008, the availability of online credit data and the abundance of processing power available, allowed 1970's theoretical machine learning algorithms (ANN et al) to change the shape of consumer lending across the globe. Real-time money transfer and credit became a reality. More sophisticated algorithms are constantly emerging in tandem with new cloud-based architectures designed specifically to open up new vistas of performance.

This massive increase in the speed of processing was a boon to consumers -- paper based processes that used to take weeks became 15-minute jobs on the internet. A series of unobservable manual processes became an integrated application process where outcomes were delivered in real-time and cash remitted in minutes. Capital One were exponents and their exponential growth was a direct consequence of this capability.

In essence though, computers were doing what they'd always been best at -- computing! Large scale arithmetic, crunching vast probabilistic models to make reasonable guesses about the likelihood of a given customer repaying or of a transaction being fraudulent. Unquestionably a step forward for industry, but the real consumer benefit was yet to fully manifest itself.


Machines are extremely good at taking immense, complex datasets and making them more easily understood. (Hello? Google). This is particularly true in FinTech where the consumer is often just seeing the very top of a veritable iceberg of service providers, payment rails, data providers, currency hedges etc.

These systems are often too complex for even an expert in the field to hold in their mind, let alone a customer without a great deal of interest. To put it another way, your smartphone can hold about 900,000 pages of text in its short-term memory, while you'd probably struggle to memorise a ten digit number in a noisy environment.

Consequently, in financial services, which tends to contain more complexity and moving parts than many industries, machines play a key role in abstracting away complexity and allowing consumers to simply do the things that they need to do. A customer doesn't need to see a table of 1,000 credit cards, they just need to see the cheapest that they are eligible for. Currently AI excels in pattern recognition and prediction, but customers want technology to play a more active role in their financial lives. Roll on...


All of the above requires the customer to provide the initial impulse to act, to search, to aggregate. Consumers don't relish shopping for insurance or figuring out who to re-mortgage with. The next wave of AI in this space will look at how to anticipate the changing financial needs of a customer, and pro-actively search out the most relevant products. In the US, Credit Karma is already playing this role in the credit space for more than 50m consumers.

People's financial lives are a complex web of interdependencies and shifting underlying requirements. As we are able to assemble more and more detailed observations of the complete picture of our financial lives, AI will allow us to anticipate our future needs, do the arithmetic on what is best for us and abstract us from the process entirely.

To some this might seem a scary idea -- ceding control to an algorithm we can neither touch or question, but this is another place where AI is already changing customer behaviour. As intelligent chat bots increase their natural language processing capabilities, they become more capable of explaining themselves clearly, and both asking and anticipating the questions we might want to ask them.

Few, if any, of us would rather spend our time shopping for mortgages when we could be sitting in the park on a sunny day. Sure, it's possible that omniscient uberminds will eventually turn us all into human batteries, but in the meantime, AI offers us the most precious commodity of all -- time -- to spend on more important things than financial services.