A friend of mine (it’s not me, honest) recently had a really bad experience with their UK bank when trying to book a flight with a top-end credit card.
This elite status card comes in a lustrous matt-finish with colored side etching to make the user (really, it’s not me, I promise) feel special about the fact that they’re spending upward of US$500 a year for the privilege of owning it.
Yet despite all of its special status and pre-approved spending power, this friend of mine still found it really hard to do something that could be made so much more straightforward if banks started to apply more algorithmically-charged Artificial Intelligence (AI) to the way they process transactions.
A problem described
Imagine someone who spends several thousand pounds, dollars, and pence on airline flights every year. Imagine if the same person has a preferred airline and so typically always uses one single dedicated website to purchase tickets. Imagine if that same person always paid off their credit card bill at the exact point of purchase, rather than waiting for a monthly statement.
You would have thought that that would be enough of a pattern, right?
It turns out that it’s not enough and the reason is largely down to external factors, however much you might think a stable pattern of behavior is enough to demarcate out who is likely to represent a risk and who isn’t.
External factors, in this case, were the recent British Airways data breach.
According to the BBC, British Airways (BA) has said that personal and financial details of customers making or changing bookings had been compromised — and about 380,000 transactions were affected.
The BA breach caused a sort of ‘industry shockwave’ that rebounded around the airline industry and caused banks connecting to booking systems to go onto hyper-skeptical alert in the days and weeks that followed.
Even a well-defined spending pattern is not enough for a bank to think it knows enough about what you are doing in order that it can keep your particulars safe, for now at least.
The additional dose of AI we need here is algorithmically programmed intelligence that ties spending patterns and payment trends to individual users.
Once that data point is created, it could then be cross-referenced with other external information including destinations purchased, other goods and services typically purchased and more.
At that point, we could also bring in external data relating to events like the BA breach, but we could appropriately ‘weight’ that data in relation to each user.
At the moment and as we stand in 2018, we can’t do this.
Wider AI application
Why should banks be spending money developing this kind of intelligence? Well because a) my friend would have been able to complete the booking more fluidly and b) banks could then start to use this greater level of know-how to cross-sell us more related products and services depending upon a level that we the users would agree to.
There’s now a call for more AI in banking at the front-, middle- and back office. We can apply this intelligence throughout.
The next decade will welcome these developments and they will be integrated with much smarter digital chatbot assistants capable of interacting with our spoken word through sophisticated Natural Language Understanding (NLU) software.
Longer term banking AI could help us eradicate fraud, money laundering and even be applied to aspects of societal importance such as transactions related to human trafficking.
My friend did get the flight albeit at a higher cost and had to endure an hour on the phone to the bank’s complaints department in order to get moderate compensation.
When it comes to banking AI, we need to explore a new world and go west, naturally.
24 September 2020
16 September 2020
16 September 2020