So far AI is being used in virtually every sector. Still, its full potential has not been realised.
Like in all sectors, AI technologies are permeating financial services around the world. It seems that following the premise: “maximizing the benefits while minimizing the risks” is not that simple.
Though, banks have continually adapted the latest technology innovations to redefine themselves as the AI-first institutions, there is a long way to go. And it is usually the case that many banks are struggling to move from experimentation around select use cases to scaling AI technologies across the organization. Then, why is this so? Is there a “recipe” for adopting AI technologies as the foundation for new value propositions?
This article portrays the driving forces and challenges to guide bank leaders discover an insight to creatively meet increasing demands for scalability, flexibility and speed that today characterise digital-native companies.
In order to address this core issue, first of all, we will categorise and describe the main “drivers” into three big headings: re-imagining engagement, facilitating cost-efficiency and de-risking risk management.
Drivers
• Re-imaging engagement: “anticipation, personalisation and differentiation”
As digitalisation is advancing in the financial ecosystem, customers are “raising the bar” developing higher expectations in the demand of performance, personalisation, quality and services.
Giving customers a more integrated banking experience across business applications is key. By aggregating data from separate sources, banks will be in a better position to proactively offer personalised products to suit customers’ needs as and when required, also adjusting to the customer’s life stages and, in that way, easy the journey within and beyond bank channels.
Understanding their customers’ behaviour, preferences and product needs banks will raise their competitiveness. Banks will require a clear strategy, that is, to adopt a design-thinking lens as they build experiences within and beyond the bank’s platform.
• Digitalization and virtual banks: Facilitating cost-efficiency
In an ever-changing market competition, it is crucial for banks to adopt a more economical and lower risk approach that will enhance their cost efficiency and overall profitability.
An intelligent infrastructure might be the answer: cloud-based platforms allow for the higher scalability while reducing costs for IT maintenance, which -in turn- enables self-serve models for development teams, which means rapid innovation cycles by providing managed services.
• De-risking Risk management
In a complex and fast-moving environment, banks need to adopt a new approach towards risk management. Build AI models that are consistent with the company’s values and risk appetite might be challenging but also an “enabler”. Early solution-ideation process to understand the potential risks and the controls to mitigate them reduce costly delays by embedding risk identification and assessment. Of course, in practice it means creating a detailed control framework that sufficiently covers all these different risks, which is a granular exercise.
Barriers
Identifying and addressing barriers to the deployment and scalability to become the AI-first will be critical. We will focus only in four threats but there are many more.
• The black -box threat: lack of explainability and poor human judgment
Explainable AI: Turning a “black box” into a “clear box”. The importance of explainability as a concept has been reflected in legal and ethical guidelines for data and ML. Articles 13-15 of the European General Data Protection Regulation (GDPR) require that data subjects have access to “meaningful information about the logic involved, as well as the significance and the envisaged consequences of such processing for the data subject.”
Banks should ensure an appropriate level of explainability of AI models to all relevant parties. Validation and explanation of how the AI arrives at its recommendations or decisions as soon as possible is a must. Still, for financial institutions and the banking system major challenges are precisely when it comes to validating the fairness and accuracy of their AI models.
• Lack of resources translated in “talent gap”
Recruiting talent is not an easy journey. Some banks have chosen to outsource or develop other collaborative models to address the talent gap. Others opted for hiring inexperienced individuals and train them, which, in the end, turns to be risky and time consuming.
Partnerships are the most common approach, which involves external consultants, innovation hubs, incubators and academic institutions. A suitable solution that brings other concerns: proprietary vs. open-source algorithms, cost of maintenance and intellectual property rights.
• Data availability and quality
Financial institutions have a lot to gain from overcoming their information readiness challenges. By creating a complete view across their structured and unstructured data, institutions will be able to analyse, understand and manage their digital ecosystem more efficiently. Achieving this comprehensive view across all their data is the first step. Only then will financial institutions be able to apply advanced AI-powered analytics across their data to extract real-time insights and increase automation to drive operational efficiencies, maximise revenue and create a more personalised customer experience.
• Regulatory framework: Data privacy and protection requirements.
When deciding on automation, everything is subject to regulation. Compliance is a must. Banks are subject to meet regulatory requirements when gathering customer data, classification of customer segments and risk profiling (see
EU’s General Data Protection Regulation (GDPR) May 2018). It is crucial to ensure that proper safeguards are in place for the AI model to make objective recommendations.
In addition to privacy, there are ethical concerns to protect individuals’ freedoms and rights in terms of choices. The ‘
Ethics Guidelines for Trustworthy AI’ (2019) note that AI should be lawful, ethical and robust.
In the end, ensuring adoption of AI technologies to become an AI-first institution is no longer a choice, but a strategic imperative.