AI use cases for the banking sector: beyond chatbots
Rudradeb Mitra, author of Creating Value with AI, says banks should look beyond data and focus on creating value to get the best from artificial intelligence.
Over the past year, I had the opportunity to interact with various bankers from around the world, and in every case, the most common question was ‘What are the use cases of artificial intelligence (AI)/machine learning (ML) in the banking industry?’
To help understand the potential use cases, here are four stories based on my experience and lessons learned during six years of working with the banking sector on AI.
Risk averseness in the banking sector
In 2005 I was working in a big Swiss insurance company and we were building an AI (ontology-driven) application to be used by brokers providing non-life insurance. What we were able to build in three years was amazing – representing all the domain knowledge in a language separate from the code and a ‘robot’ which understood (or compiled) the knowledge to generate the application. This architecture made sure that the application would not break if the application requirements were changed or new requirements were added. It was one of the first AI applications successfully built in the banking sector. However, despite achieving technical success, we did not achieve business success as the application was not deployed in production.
One of the things that I learned from the three years’ experience is that the banking sector is highly risk-averse and everything is driven by return on investment (ROI). The desire to use new innovative technology is very low. The decision-makers in the banks often lack the technical expertise required to make those ‘brave’ decisions and are mostly driven by job security.
ROI may not always be important
My next stint with the banking and insurance sector was in 2013, when we were building our insurance tech startup. Despite being part of a bank’s incubator, the adoption was painfully slow. However, that all changed when we were able to show that we have active users and have collected real data, something that the bank’s digital products were unable to achieve.
I learned that banks often struggle with user adoption of their digital products. So, if I am not able to show monetary ROI, I will try to show the value of a product or technology in other aspects like improving user engagement or building trust.
Improving customer experience is key
I had the opportunity to listen to bankers from all over the world (South America, Europe and Asia) at the Efma CCX Conference. I repeatedly heard how important it is for them to improve customer experience and regain trust. It seemed to be one of the biggest focus areas, although most of the ‘AI’ for customer experience was limited to chatbots.
The lesson learned from this was that improving customer experience (both online and offline) seems a key challenge that banks are struggling with. If you focus on products that can improve customer experience, banks are willing to listen. But chatbots can only improve one aspect of the customer experience.
Banks want to adopt new technologies
Recently I was in Kuwait, where I spent time with the digital and back-end team as well as senior executives. One question that was often repeated was: ‘Why is there so much focus on AI Now? Is this all hype or is there some substance? After all, it is not a new technology.’ My answer was that whether it is a new or old technology is not relevant. The question should be whether the technology is able to help me solve a problem that didn’t exist before.
Throughout the world, user behavior is changing, and if banks want to be ahead, they need to adapt to new user behavior and expectations. ML and AI can help them build products to meet the demands of the new age.
I learned that bankers do realize that their customers and users are changing their behavior rapidly, and they want to build intelligence in their processes to adapt to that. But many are unable to connect the dots and create a full circle, to see how AI/ML can help build products to improve customer experience.
When I bring all the lessons together I realize that if banks start thinking of ROI for AI products, they won’t end up doing anything. Innovative products often do not clearly show the ROI. But if banks start thinking from the perspective of value creation for their customers, there are a lot of use cases of AI/ML.
Here is how banks need to work on AI and ML.
Begin by listing broken processes or desired processes
There are many processes that banks can fix or improve, whether in the back end, mobile customization or in the front end. A lot of processes in the banking sector were made within the limitations of the human mind. There is room to improve the processes by combining the strengths of human and machine intelligence.
The most suitable use cases that can be solved by a combination of human and machine intelligence include those with distributed data; high quantifiable patterns; where there is high human error with the existing process; where large amounts of data can be generated or collected; and where there are low to medium intuitive patterns or the need for emotional intelligence in parts of the task, which cannot be quantified.
Think from the product point of view
AI is not rocket science or some magic to fix all problems. But think from the point of view of building products using the power of AI/ML that can fix or build new processes. Do not ask: ‘How can I use AI/ML?’ Instead, ask: ‘What products can be built to solve the problems?’
Banks have a lot of money, and often they try to think big – they want to build a big team and create a perfect solution. Unfortunately, that is not the way to go ahead. Start small and instead of building a big team and a big vision, start with a small team building proof of concepts. Keep the cost low. Sometimes it is good to be risk-averse, but not so much as to not do anything.
Focus on creating value, not on ROI or data
Please do not start from data. So many times, I have heard people saying: ‘We have a lot of data; let’s see what we can do’. I am sorry, but this is not the right approach. It will be like looking for a needle in a haystack. Instead, focus on creating value for your users.
ML can help banks to achieve excellence in a wide range of key focus areas for operational excellence. It can be used to help the organization increase transparency and frequency in communication, focus on easy-to-access and easy-to-deal-with private bankers and simplify process, reduce the number of relationships per relationship manager, and enhance the technical and product skill set of private bankers. ML is especially relevant in helping banks to refocus on the client risk/return profile, enhance proactiveness and anticipation of client needs, simplify products, enhance the array of products, and produce more reliable, timely management information and comprehensive external reporting.