Here are 4 ways AI is streamlining banking in India

India is estimated to have the highest number of digital banking users in the world, as of 2022. The country also has the most diverse and dynamic customer base in the world, making inclusive financial growth necessary.

Today’s customers can access various customized services and products at their fingertips, and banks are not exempt from these expectations. Competitive pressures also force banks to focus on improving productivity and deal with the ever-increasing risk management and compliance demands.

The India stack, a set of open APIs (application programming interface) for government, businesses and individuals to utilize India's digital infrastructure, has deepened and widened the access to financial services in what was traditionally a cash-driven economy. India's unique ID programme, Aadhaar, a consent-driven architecture where the unique ID/number can only be verified with the cardholder's consent, has dramatically lowered the cost of confirming customer identity. For example, the population covered with bank accounts increased from 53% in 2015 to 78% in 2021.

All this has created an ecosystem which has made financial services truly digital.

This has also paved the way for technology, especially artificial intelligence (AI, to drive innovations in unforeseen ways. AI provides banks an opportunity to support improved delivery of banking services and products.

Moreover, generative AI’s distinguishing ability to analyze large datasets and create content in consumable formats will be a game-changer for the banking industry. Globally, banks have been using statistical models to evaluate the creditworthiness of customers for decades. Hence, it is no surprise that banking is the leading industry when it comes to the adoption of AI in many different areas:

Retail credit has been India's largest area of credit deployment, and it continues to grow rapidly.

Increased volumes and velocity of loan applications have necessitated shortened risk assessment and evaluation cycles while ensuring better risk management and compliance.

Underwriting is a complex process that requires adherence to bank policy, regulatory compliance, risk assessment and customer due diligence. AI algorithms are being used to assess creditworthiness by assessing alternative data sources, such as regular utility payments and consumption patterns. Generative AI further allows the assessment of unstructured datasets, enabling underwriters to make informed and accurate credit decisions.

India has seen a steady rise in investor accounts over the last decade. The average Asset Under Management (AUM) of individual investors has increased by 21% as of June 2023. However, India’s AUM as a percentage of its GDP stands at 16.9%, vis-à-vis China at 20% and South Africa at 62%.

Banks are uniquely placed to capitalize on the growth in retail, considering India has the second highest number of high net-worth individuals amongst BRICS countries.

All these factors have created a unique opportunity for Indian banks to step up and cater to the proliferating needs of both new and existing customers.

Banks now have the ability to provide tailored investment and algorithmic portfolio management through enhanced usage of artificial intelligence. For wealthy customers, banks can use generative AI to analyze data on companies, summarize financial reports, analyze a customer’s portfolio, and make investment recommendations based on the individual’s risk appetite, financial goals, etc. Banks will be able to leverage AI to create portfolios for retail investors who are looking for safer investments. Hyper-personalization at an individual level, at scale, with a high level of consistency, has been made possible by AI.

Digital banking has democratized access to financial products for customers, leading to increased risks banks face. Risk management is becoming complex due to the volume and variety of risks. In addition, their scale and scope continue to increase due to the growth in product services and channels through which they are distributed.

Traditional AI combined with generative AI has the potential to add tremendous value to the process of identifying, assessing, and mitigating risks faced by banks. One of the challenges is the risk of the increasing number of false positives – where transactions are wrongly flagged as suspect – overwhelming areas such as transaction monitoring and fraud. For example, every transaction that is identified as fraudulent or a potential money-laundering violation needs to be triaged, verified and flagged; a process that at great volumes can overwhelm systems. AI will help risk management teams identify multiple types of risks at scale.

Banks continue to invest in technology to automate compliance processes as they are complex and time-consuming. The scale of compliance activities continues to increase in terms of volume and velocity due to the growth in the number of transactions and the channels through which they take place. Generative AI will assist banks with critical capabilities to manage this scale.

Its algorithms can ingest large amounts of data, analyze regulatory changes and generate risk assessment drafts using this information. In addition, AI can draft responses to queries from internal stakeholders and regulatory authorities.

Banks have been expanding their digital banking services over the past decade. However, until the advent of generative AI, most of these services required human presence to advise and guide customers. For example, chatbots were not fully conversational and had limited functionality. Changes to the underlying knowledge base required retraining of existing AI models. Generative AI has now addressed these issues.

With greater training and inferencing capabilities, generative AI will be able to apply knowledge and patterns in new interactions that will be of immense use in customer service. These capabilities are not just limited to customer service but open up a whole slew of possibilities in the area of conversational banking.

Almost all AI systems today can be categorized as "System 1 AI," an AI system that synchronously responds to queries with an answer. For example, when a customer wants information on a credit card while conversing with a chatbot, the chatbot analyzes the query and responds with the best answer from its knowledge repository.

AI systems have traditionally lacked the ability to build a problem-solving approach, invoke reasoning and allocate time and effort to solve a problem by following a stepwise approach (System 2 AI). You cannot ask an AI to “Create an annual report” and expect it to figure out all the data sources, fetch the data, create the P/L, generate the text, and assemble the final report.

Since the advent of generative AI, this has changed. Some AI models show glimpses of how System 2 AI can function. Advances in AI (specifically generative AI) are allowing the integration of System 2 mechanisms with well-evolved foundations of System 1 AI. Several enterprises, including Axis Bank, are conducting experiments on building AI platforms that can harness the rational decision-making ability of System 2 AI in combination with System 1 to solve problems beyond the realm of System 1 AI.

At the same time as these advances, we must not lose sight of the ethical dimension. Banking is a highly regulated industry; hence, transparency, data security, accountability and unbiased decision-making are critical. Banks’ AI usage should incorporate these key areas holistically.

The explainability of AI models is a crucial aspect, as it is important to know the key drivers behind the decisions arrived at by these models. We need to ensure adherence to data privacy and security while using AI models. There are risks that they may exhibit certain biases inherent in training data or leak sensitive data. Bias in AI algorithms can lead to incorrect decisions or, worse, may end in upholding racial, gender and other discriminations. Accountability for wrong decisions made by banks due to algorithms is another complex issue since these algorithms are primarily opaque.

It is not often that technology comes along and simultaneously changes how society has been thinking and functioning. It is equally rare to have a present as exciting as the future. And AI is making exactly that happen.