Author: Mr. Bálint Bakó, Product owner of AI team at ApPello
The opportunities offered by artificial intelligence (AI) are extremely attractive in numerous fields, banking included. Interest in AI is growing exponentially and the application of more advanced tools provides countless advantages. There is simply no question that this is the future, but how can a few shoots of development become a complex ecosystem?
Just as the first industrial revolution replaced the physical strength of workers with machines, in the fourth, ongoing industrial revolution, artificial intelligence is similarly reshaping the world in many sectors including healthcare, marketing, e-commerce and finance.
In recent years, decision-makers in the finance sector have also turned their attention to data-based decision-making since the technology holds enormous promise, whether it be for business development, customer experience or risk management. There is broad consensus that the large reservoir of data available to banks is well worth exploiting with tools that are more advanced than we have today because one of the by-products of the operation of banks is the vast resource of ‘well-structured’ numerical data. Today, plenty of suppliers offer narrow AI applications, but at the same time these products generally only represent a solution to some particular range of problems and they make little use of locally available data assets.
Of the financial institutions participating in the McKinsey survey, 60% employ at least one AI-based solution. For global banking, they estimate that AI technologies could potentially deliver up to $1 trillion of additional value each year. This mind-boggling sum could derive from growth in sales, cost savings and exploitation of until now hidden business opportunities. However, on the basis of the survey by the consulting firm it is evident that many banks are struggling with difficulties when trying to shift from isolated, experimental use to solutions that encompass the entire organization.
One common misconception about using AI in banking is that – having an awareness of the full profile of customers and being able to ‘work out’ all their thoughts – it will replace customer management in its entirety. This is far from reality; instead, in the foreseeable future, AI is likely to be used effectively ‘for the time being’ to support customer actions and the work of staff.
The first harvest
In the wake of early applications, artificial intelligence has attained a role in numerous areas of the finance sector. The most common areas of application are the automation of processes, virtual assistants and machine learning employed in risk management.
As a generality, one can state about the generation born around the turn of the millennium that, being fully cognisant of the technological possibilities that are available, they have no interest in queuing at their local bank, waiting to be connected on a helpline or filling out reams of paper to access a single financial product. Instead of using traditional, long lasting credit decisioning process, AI is able to provide more accurate decisions, based on transactional data and other behaviour data, which take into account potentially several tens of thousands of variables. These systems can provide lightning-fast and paperless credit assessments even for students or the self-employed who, with traditional credentials, would have limited access to financial resources.
Every financial service provider finds customer service activities a notoriously resource-intensive area. Thanks to automated systems based on artificial intelligence, this area may require far fewer employees, in addition to which a fast and accurate service can serve to improve the customer experience and thereby even increase sales.
How to avoid this turning into a jungle
However, as is the case with all disruptive technologies, the future raises a number of questions. Will it be to our advantage if business decisions are made exclusively by machines? In the end, will humans be left out of the process all together? Is it sufficient to synthesize processes and reach decisions based solely on historical data and existing data assets? Would a machine decision-maker make fresh capital available for a previously unknown renewable energy innovation based solely on historical data? Would the descendants of previously blacklisted social groups receive personal loans?
Today’s technology is already capable of defeating chess grandmasters in any game. But would AI be able to develop a next generation chess program? For a moment, let’s set our imagination free and adjust the analogy to a banking environment: how would this work in a decades-long simultaneous game where the rules are different on each board, where the opponents are switched around and new games are constantly added to the board?
Many are concerned that, without the strict outlining of frames, the genie will escape from the bottle. Research conducted by KPMG on the subject found that regulation of the use of artificial intelligence currently lags behind its market penetration, although in this respect practices differ from country to country. 87% of IT managers surveyed said regulation was absolutely necessary, yet an even higher proportion considered that companies needed to pay more attention to corporate responsibility and ethical considerations. A critical point in systems built on utilization of customer data is data protection. Executives involved in the survey consider this area will be particularly important in regulation.
The garden thrives in the hands of the caring owner
So how will IT expertise and professional banking creativity find each other? Where is it worth starting? The answer lies in an institution’s own data assets. Financial institutions often choose the more complex path, wanting to build on external alternative data resources since they see this as giving them a competitive edge and they tend to forget the inherent potential of data already at their disposal. Naturally, it is also a legal question whether the primary data of customers can be used safely for forecasting and automation, or whether processes are developed based solely on imported, secondary findings from specially conducted ‘laboratory’ analyses. The situation is similar to establishing an arboretum: one may plant a varied stock but in the end one cannot override the climatic conditions of the given region.
Progress is frequently hindered not only by legal regulations but also by the mindset of organizational departments: people jealously guard their work and stick to their best practices because they do not know how the new technology will affect their everyday lives, which is why they prefer to slow down its expansion. They often forget that it could rid them of monotonous manual and error-prone tasks, freeing up valuable time to work creatively, identifying business correlations which create genuine added value for companies. At times like this, there is especial need for a team of experts capable of coordinating business requirements with the available new technologies and pinpointing solutions that fit the size and characteristics of the customer base. Thus, one of the most important steps in exploiting AI is to have sufficient creativity within the banking organization to articulate needs that previously seemed unfeasible. It is possible that a trained team of developers, leveraging AI, can come up with a solution.
We at ApPello have trained our team in this approach. We move from simpler solutions to more complex ones, always bearing in mind feasibility, customer satisfaction and last but not least, cost efficiency.