Dr. Adrienne Heinrich

Head – Artificial Intelligence (AI) Center of Excellence, Union Bank of the Philippines

Good credit scores are hard to achieve and worse, loans are hard to obtain in the Philippines even today. Bangko Sentral ng Pilipinas (the central bank of the Philippines) indicates that approximately 44% of the country’s adult population is unbanked. This hinders many from establishing businesses and discovering growth. AI is changing this, improving fair economic welfare with models to help assess leasing customers who lack traditional data by using alternative data, resulting in favourable credit scores for part of the population. This will help customers attain loans where the risk of default is low.

With each bank offering the same products and services, the room to compete is scarce. Hence, one of the true advantages that banks have over their competitors is customer satisfaction and addressing the evolving needs of the customers. This is why a great customer experience is paramount for a bank, and AI will help achieve this, elevating them to new heights to reach greatness and stand out from the competition.

AI use cases in the BFSI industry:

Digitalisation, end-to-end banking and financial processes are avenues in which AI can further accelerate the BFSI industry. Today, it extends to more than mere chatbot implementation and automation, evolving the BFSI industry, by promoting financial inclusion and improving customer satisfaction.

• #1 Lending credibility

Gender imbalance is an ongoing issue in credit lending. Men frequently attain loans more than women based on the biased assumption that men will often be able to pay them back. This bias is encoded in various non-gender data fields and cannot be easily removed by prohibiting the use of gender data information. Combining responsible access to gender data with advanced machine- learning techniques can help significantly reduce the discrimination women face. This can be achieved by approving loans for customers who deserve financial support but are currently discriminated against when traditional ML model approaches or regulatory binding guidelines are applied.

A study by the World Bank highlights that increasing women’s access to credit not only improves gender equality but also helps reduce extreme poverty simultaneously. On a large public dataset, it was found that organisations will also benefit from increased gender equality, as also their profitability increases, resulting in a win-win situation for everybody.

Combining responsible access to gender data with advanced machine-learning techniques can help significantly reduce the discrimination women face.

• #2 Explainable AI

The trend that we are observing is that the better performing AI models are in general the more complex ones (e.g., deep neural networks) that are not so easily explained nor transparent. With the penetration of AI and digital, we must be proactive in mitigating potential risks regarding ethics and trusting AI outputs. Therefore, more attention shall be put to developing explainable and responsible AI (XRAI) so that both the data scientists and the decision-makers know when to trust that the AI output is reliable and fair. With a set of XRAI tools and frameworks, we want to get to the level where we understand the main patterns and criteria an AI model follows, how the AI model decides in selected individual cases and what kind of biases are present. Just like when you are interviewing a human being to find out whether this person is fit for the job, you now need to understand whether the AI model is fit for its job. With this in hand, the stakeholders will make better decisions for both the customers and the companies. Explainable and easy to understand visualisations of the AI output can present an opportunity for decision makers to learn so far unknown patterns in big data and even correct for potential bias on the decision maker’s side. This results in a more efficient workforce, ensuring banks expertly leverage their resources and time to solve complex issues by optimizing the collaboration between humans and AI.

• #3 Customer satisfaction management

Customer feedback is invaluable to improving banks’ products or services. Ratings and reviews are crucial, helping banks determine what works and what does not. However, an influx of descriptive reviews can prove tedious for banks to sift through for information. Hence, AI and data science help ease this problem with natural language processing (NLP) capabilities.

NLP models will sift through the feedback, containing thousands of text excerpts to underline customers’ key concerns and issues. The detailed analysis can help banks better target customer groups most important to their business strategy – by having a different action plan for gaining new acquisitions than for maintaining existing customers. AI is necessary, as it allows banks to efficiently determine the different issues that various customer groups face on the same product in a shorter period. In addition to operational efficiency, this process change will eliminate subjectivity and inconsistencies across ‘readers’ and allow for improved prioritisation and informed decision-making. Because of it, banks can better comprehend the customers’ situation and sentiments, implementing better improvements for their products and services.

Gone are the day’s AI was used merely for chatbot implementation, with its usefulness proving vital to reducing discrimination, enhancing decision-making and improving customer satisfaction. One thing is clear – AI doesn’t look to slow down, and banks looking to deliver the best customer experience should take note of its prominence.