Bank customers are becoming more and more demanding. In the age of Google, Apple, Facebook and Amazon, we have become accustomed to personalized offers building on data that we have voluntarily provided. As a result, the first banks have already expanded on the AI system used by Alexa, to produce finance-specific answers for customers who want Alexa to run their investment strategy.
The rise of machine learning algorithms
Behind this development are machine learning algorithms, which are able to model the characteristics of the people concerned and predict their preferred investment behaviour and interests. While these algorithms can learn, the “machine” element does not make them self-sufficient and self-sustaining. They must be fed the right models at the right intervals by a human being, in this case a data scientist.
This is by no means the only use of AI and analytics in banks. One of the most common uses is in managing unstructured data, including emails, news articles, excerpts from the commercial register and recorded telephone conversations. Analysing these data does not necessarily require AI; the first challenge is to be able to apply analytics at all. AI, however, certainly gives the best chance of processing this information intelligently.
Digitalization in banking needs to be performed from end-to-end, encompassing activities in the front-, middle-, and back-office, and underpinned by advanced analytical technologies. Whether you are a business sponsor, data scientist, data manager or IT facilitator, our practitioners have the experience and expertise to help you navigate the analytics path to value. Sign up for our monthly webinars and accelerate your analytics journey.
Smarter banks with AI
AI also has potential to make banks smarter. For example, it could be used to learn how markets behave, through deeper insights into the behaviour of the market participants, enabling better risk assessments. Modelling human behaviour — complex, emotional and influenced by a wide range of inputs — can also help bankers. An AI system that has learned the behaviour of a trader and its effects on performance over time may help to prevent that trader from making unsuccessful decisions based on ‘gut feeling’.
However, it is also conceivable that machines themselves learn by “observing” successful human action over a period of time and then modelling it mathematically. Again, this is likely to need a data scientist to develop the correct mathematical model and manage its adjustment over time, building a long-term partnership of mutual learning and adjustment.
AI also makes the bank smarter than its customers, so that it can offer more useful services. This can happen in several ways. AI can aggregate all information about a customer, so that it ‘knows’ the customer, and can tailor its interactions. It is also conceivable that Apple’s face recognition software could play a role. The bank branch of the future may ‘recognise’ me as I walk in through the door, so that the consultant who greets me already knows about me.
Some are sceptical: One analyst recently commented to me: “What good is knowledge about a customer at the entrance, if the consultant is getting a coffee at the time?” There are, clearly, issues about the efficient operationalisation of the knowledge gained from AI.
AI in credit scoring
In 1970, Irish banks closed for six months because of a strike. There was, relatively quickly, a shortage of cash. But the economy did not grind to a halt; far from it. Instead, people began to exchange cheques and other instruments in the pubs and stores, creating a kind of alternative currency. They could only do this because there was mutual trust in the local economies and within communities. Everyone knew who had exchanged what and who was trustworthy.
In other words, reputation mattered, and turns out to be crucial in credit scoring of any kind. It is already possible to obtain relevant information from data that people have generated online. Examples are social media posts, browser behaviour, phone calls or online payment histories. This enables people to build reputations in regional or global communities. This data use may seem a touch obtrusive. However, for people in emerging markets or developing countries who do not have access to traditional banking, this may be the only way to build a reputation and a credit history.
AI as weapon in the fight for survival
Since the global economic crisis in 2008 and regulations from Basel to Sarbanes-Oxley, cost pressures have been accumulating on banks. It has also created enormous potential for disruption. Now the question is whether the worm will turn, or if fintechs will become market leaders. Today, fintechs and banks are predominantly complementary. It remains to be seen how quickly the big players will realise the opportunities of digitalisation — that is, not simply replacing analogue processes with digital, but discovering completely new potential in datasets and AI.
The biggest challenge is probably cultural. AI needs an approach of ‘fail fast, fail quick’, but banks still find it hard to accept failure. With RPA and AI, the financial world now has a way to give employees the freedom to start this cultural change. Go on, dare to take that first step.#AI in banks can be used to learn how markets behave and enable better risk assessments. Click To Tweet
Enterprise readiness for AI
SAS recently conducted 100 interviews with business leaders to understand the current state of AI readiness in corporations. You can read the full SAS report here.
This blog post was first published in German on the regional SAS blog Mehr Wissen.