There has been a lot of talk about AI powering ‘roboadvice’ and predicting markets but Artificial Intelligence (AI) has the potential to dramatically improve retail banking for consumers as well - detecting fraudulent transactions and anomalies, predicting behavioural patterns, personalising service and simply working out the best things to do with your money. In the second in our series of articles demystifying AI, we take a look at what makes AI and Financial services such a great combination.
Common Sense Takes Time
Opening up the data
Artificial intelligence needs data - LOTS of it (after all, it’s machine ‘learning’ not ‘machine learned’ - ‘common sense’ takes time….). Historically access to data was strictly limited to the big banks, where vast amounts of data were housed, hoarded, and largely left alone. However, there is a global shift towards open data - sharing data between third parties with users’ consent - as facilitated by cloud computing and open API’s.
Open banking opening doors for innovation
Open data in financial terms essentially is ‘open banking’. Already initiatives coming from PSD2 (the new open-banking regulations) are making significant headway in the UK and Australia is following suit. The new directives will have enormous implications for banks and fintech startups where access to data opens doors to build personalised, pre-emptive and predictive services for consumers. And we’ll see collaborative models emerge where banks are able to offer hyper-targeted and enriched services to their customers.
Generating data today, like there’s no tomorrow.
Machine Learning a more technical term that encompasses most of what we would consider to be modern AI - this is totally reliant on data. Deep Neural Networks (one class of machine learning) needs sometimes hundreds of thousands of training examples to be able to even tell the difference between a cat or not a cat (Fun link for people interested in AI history).
Fortunately for cats, approximately 2.5 Quillion bytes of data is created, processed and stored each day. The emergence of the ‘Internet of Things’ (IoT)- being sensors and devices (from your fitbit to your fridge) that autonomously capture behaviour and share data with the internet is propelling this. Not only will businesses have more data, IoT promises to deliver better data as well. With Open Banking, IoT and cloud computing it is fair to assume that there will be an explosion of AI in financial services (as will the general explosion in AI as discussed in our previous article).
Enter your email below and you'll be the first to know when Xinja is ready to launch.
Fundamental to a Frictionless Future
What does all this mean for your banking?
No longer will the customer have to stand by and accept the irrelevant FAQ, or wait for hours for the relevant topic to uttered from an interminable list as they wait on the phone. Technology will make life easier.
Chatbots and virtual assistants are fundamental to a frictionless future in financial services. The ability to respond to any number of customer request at virtually any time with lightning speed is an exciting prospect for consumers.
Furthermore, these bots will also be channel agnostic - meaning you could get your google home to pay that utilities bill, or send an emoji on Facebook to split a bill, and take a selfie to get a mortgage...actually maybe not that last one but you get the point. For simple requests chatbots and VAs have got you covered. The problem though is that this technology at its current state doesn’t have as much common sense as we would like.
When things get a little more complex, inquiries will be immediately transferred to a friendlier and more skilled service agent. What’s cool about this is that you will only need to put your information in once and the AI will make sure the right information is put straight in front of the service agent, enabling faster and more personalised service straight off the bat, if and when your chatbot is not doing the trick.
Context is the key
In the medium-term AI will be able to predict your questions and return answers before you even ask. The mighty FAQ page might find itself redundant as hundreds of more nuanced personalised responses begin to replace existing systems. Monzo - a fellow neobank in the UK (one of whose founders, Jason Bates, is on the board of Xinja) - with their smart FAQs has been experimenting with machine learning to take into account a whole range of user interactions and return FAQs that match.
Knowing what we need before we do
But it’s not just customer service that will benefit. Overtime, with AI, basic budgeting can build towards prediction of spend, taking into account temporal relationships (basically looking at transactions and other external factors, such as the time of year or month, purchases made when an account balance reaches a certain level, or when purchases happen in reaction to others.) Then virtual financial coaches can pre-empt, anticipate and nudge appropriately, because they understand the context and can act according to the preferences and goals the individual user sets for themselves. They can promote ongoing ‘money mindfulness’ (which is what Xinja is all about).
Why not personalised mortgages?
Personalised service becomes personalised financial service
Digital has brought about the age of algorithms where AI or more specifically machine learning has personalised everything from our news feeds, what we watch, to what music we listen to and what we buy. Our point? If digital can offer services that are tailored to each unique customer why should financial services be any different? Credit scoring, for example, currently quite a blunt instrument, can become more and more nuanced to the individual a more data gets sucked in. We’re already seeing personalised insurance rates (based on how you drive/behave etc. rather than lumping you in with your broad cohort - check out Lemonade). In banking, why not personalised mortgage rates that modify automatically?
A word of caution…
Machine learning has been criticised in some cases for being a ‘black box’ and for generating its own form of bias, and risk of discrimination. However, there are lots of checks & balances that can be used to test models against this kind of thing, and overall, the outlook is good for creating fairer systems.
So is this the age of the personal banker?
There is every possibility that AI will deliver in banking what it has elsewhere - a virtuous circle of improved products and services to the individual fed by an ever increasing supply of data. In other words, every banking experience becomes a personal banking experience. And at Xinja, this is right up our alley 🙂