Artificial Intelligence (AI) is powering the banking industry’s ability to increase customer engagement and deliver exceptional customer-driven experiences. Though many banks are still in the nascent stages of application, the use of AI is growing steadily across the enterprise. AI algorithms and advanced analytics are applied to make instant decisions using real-time data. This is to perform complex and intelligent functions associated with human thinking, unlike machines that react to rules-based logic or deliver predetermined responses.
By combining and analyzing information from a variety of different sources—including sensors and remote inputs—AI can create a more efficient, informative, and secure banking experiences, delivering high value to the customer and increasing a bank’s profitability.
According to IDC, the banking industry leads the way in AI adoption, with $5.6B expected to be spent on AI-enabled solutions by 2022, including automated threat intelligence and prevention systems, and fraud analysis systems. Furthermore, a recent analysis by IHS Markit shows the actual business value of AI in banking worldwide is expected to reach $300B by 2030.
These investments are fueling rapid AI and machine learning (ML) innovation and are transforming the customer journey across the banking industry. The wider availability of powerful cloud-based AI and ML tools has dramatically lowered barriers to entry, thus changing the competitive dynamic across the industry, from retail to commercial banking. The differentiation comes down to how and where AI is applied.
A Note About Data
Successful AI application in banking begins by customers having access to easy-to-use, intelligible digital products, such as apps available through multiple interactive devices. Such products drive customer engagement, generating more data regarding preferences, usage patterns, and more. AI algorithms use this data to continuously train the models, resulting in improved products and experience. Because of the large amounts of data, the banking industry has the most to gain from AI/ML solutions; however, most of a bank’s data assets are siloed and difficult to ingest due to regulatory constraints and legacy technology stacks. It is therefore critical for banking to introduce a strong data and AI/ML strategy.
A recent investigation found widespread AI adoption in banking worldwide could save in excess of $1T by streamlining and improving front, back, and middle office functions; furthermore, another recent study, this time by Juniper Research, predicts AI-powered chatbots will help banks save more than $7B per year by 2023, representing a 3,400 percent growth in operational savings from 2019.
The majority of AI/ML applications deal with routine matters such as administrative workflow, error reduction, connected machines, and fraud detection. Contrary to common belief, cost saving does not necessarily mean banks have to replace humans with automation. In relation to client service and customer engagement, AI works with humans, carrying out back-end technical tasks and making workflows more scalable, providing assistance and recommendations for humans. This frees frontline staff to increase time spent focusing on high-value driven tasks and engage in more face-to-face contact with customers.
Banks are leveraging AI-driven applications, such as intelligent banking conversational interfaces and chatbots, to create greater efficiency. The AI-led solutions improve workflows by decreasing manual processes, such as post-incident data collection, analyzing details, identify vulnerability cases or validating loan documents.
Real-time communications enabled by AI technologies have opened multiple avenues for innovation, from remote banking advice or wealth management advisory, to facilitating online appointment requests, delivery of bank credit score results, and supporting customer messaging.
Sweden’s Swedbank partnered with an AI voice and natural language processing (NLP) company to integrate a conversational AI virtual assistant, Nina, with the bank’s contact centers. Unlike a simple FAQ list, where a set of questions and answers are provided, Nina is programmed to converse with customers in real-time relating to a variety of scenarios. Within three months of deployment, Nina handled over 30,000 conversations a month with a 78 percent “first-contact resolution”, hence saving operational costs. Banks investing in an AI-powered chatbot platform like Nina can deliver a higher degree of personalized interactions. The chatbot can converse with customers within a messaging interface and leverage machine learning to more accurately interpret questions and provide real-time answers. Customers receive answers more quickly and efficiently and banks have the added benefit of reduced operational expenses.
Xiaowei, a chatbot serving China Construction Bank’s WeChat app and mobile bank handled 1.9B customer interactions in 2018. The AI-enabled virtual assistant is programmed to understand and respond in 56 different Chinese regional dialects, and understand more than 80 types of banking services. As with all chatbots, Xiaowei provides around-the-clock assistance, even during the holidays, so customers need not wait for normal banking hours to get personalized information.
While the use of chatbots has been a major step forward in the digitization of banking, humans will always have a place in customer interactions. Modern solutions are able to perform sentiment analysis and adjust on the fly, therefore, chatbots must allow for fast and automatic transfers to an available banking agent in certain cases and scenarios. Transfer should happen within the same conversation window, meaning the transfer is communicated to the customer and the agent gets the full conversational transcript of the conversation, eliminating the need for repetition. Should a transfer be required when an agent is not available, the chatbot can set customer expectations on response times. However, this should be integrated with a workflow ticketing system that will tag, prioritize, escalate, and route the conversation to the next available agent.
Predictive Analytics Tools
The benefits of efficiency gained by applying AI extend beyond routine task management. AI-enabled financial documentation helps banks cut down on documentation time and upgrade reporting quality. This computer-assisted technology can also improve real-time customer care enquiries by ensuring customers digitally receive an accurate spend history and personalized recommendations directly onto multiple interactive devices.
In the digital era, banks must quickly and accurately analyze thousands of mined data points, including demographics, financial standing, and product history to improve personalized services using cognitive computing capabilities. This same data also informs fraud detection and can impact decisions about taking out new banking products and services by analyzing long-term patterns overlooked when making decisions on the go.
In order to process customer emails more efficiently and save the customer service team’s valuable time, UK bank NatWest has developed AI models to quickly scan and route emails according to topic and category. The bank started by building a single decision-making layer drawing on relevant customer data. This involved understanding tasks carried out by frontline workers and programming those tasks into an AI/ML enabled prototype.
Tasks range from taking an invoice and extracting information from it, to answering customers’ questions. The bank’s aim is to leverage data and decision-making capabilities in order to understand and personalize the service to each customer and conduct one-to-one conversations with them. NatWest built an email categorization algorithm for the bank’s customer service staff, as the vast majority of NatWest’s customer interactions are via phone or email.
Better Connection, Better Banking
Machine learning supplements and streamlines workflows in banking, and increasingly helps to removes frustration from the customer experience. Consider the AI-powered virtual mortgage adviser offered by SPF Private Clients as a key example of “outsourced” banking via an AI application. Buyers receive mortgage recommendations in under three minutes and are connected to a mortgage adviser in 30 minutes—a service previously taking five working days. These mortgage virtual assistants interact with property buyers and connect borrowers and lenders between office visits to reduce unnecessary costs.
In an industry tasked with keeping customer’s data and money safe, fraud detection is top of every bank’s CTO’s agenda. In commercial banking, over $10B in rogue trading losses have been incurred over the past decade from the top 13 global banks. Anomaly detection is a key approach helping banks identify fraudulent transactions and transfers, supplementing traditional rule-based approaches. With predictive analytics, banks can more easily detect fraud and score transactions based on a wide range of customer profiles and data.
AI/ML applications scroll larger databases looking for abnormalities, communicating possible future steps to the bank for prevention. Abnormalities may include differing bank sort codes or alternative customer names under the same account number. This is not only useful in recovery—the analysis of information may assist with pre-diagnosis and thus a better future banking outlook.
The Future of Banking
The availability of AI-enabled solutions allows banks to deliver customer-driven experiences, providing application strategy and methodology is planned according to business needs.
AI and ML will continue to greatly benefit the banking industry, but the technology’s full potential can only be experienced if the banking infrastructure is able to support the technology and data needs. As banks increasingly become reliant on ML to provide predictive analytics, they will need to meet new regulatory and interconnectivity demands.