By Ashvin Parekh, Managing partner, Ashvin Parekh Advisory Services LLP
When it comes to artificial intelligence (AI) adoption in the financial services (banking, financial services, and insurance) sector, India ranks as a global leader. According to some recent surveys by recognised firms, more than 30% of Indian companies have been trying to or plan to maximise value from AI. Fintechs, banking, and software sectors in India are at the forefront of this, leveraging AI for tasks such as customer interface and servicing, credit risk assessments, process and control efficiencies, and automation and fraud detection. The adoption of digital is exceptionally high, and India can boast of the highest fintech adoption. According to a survey, the adoption is 87% compared to a global average of 64%. About 46% of the world’s banking digital transactions occur in India, reflecting the country’s strong digital foundation.
There has been strong government backing as well as regulatory encouragement. The Reserve Bank of India (RBI) has played a significant role in both digital as well as AI adoption in banking. With the support of policymakers and regulators, the nation has built innovation and skilling hubs across several centres of excellence. Initiatives like India AI Mission and Digital India Bhashini are supporting indigenous AI models and language technologies, further strengthening the ecosystem for AI innovation and accessibility in financial services. There are guidelines and regulations framed by the regulators in areas like algorithmic trading, robo-advisory, and digital lending. The Digital Personal Data Protection (DPDP) Act establishes comprehensive data protection requirements for AI deployment in BFSI.
This article, in the backdrop of such an excitement around AI, examines the ways to make AI adoption in any BFSI constituents deeper and wider. It examines the essentials and parameters to build a robust framework of governance, and critical ingredients essential for a linear and structured growth of AI application in a company. Some of the pitfalls and risk management of undesired consequences are also discussed.
To help BFSI firms identify AI use cases and drive wider and deeper adoption, a company could follow these structured steps. The first critical step would be to align initiatives with business objectives (example revenue, growth, risk reduction, customer experience, etc.) followed by the creation of a framework for measuring impact from the shortlisted use cases. According to a survey report published by the Bank of England (BOE) and the Financial Conduct Authority in 2024, the use cases ranged from optimisation of internal processes to cybersecurity and fraud detection. The survey also observed that a third of all the respondents across the BFSI sector deployed third-party implementation. This proportion of third-party implementation is accepted to be higher in India’s BFSI sector, thanks to the growth of infrastructure including software engineering institutions, incubations centres, and a large number of entrepreneurial coders and solutions developers.
In regard to the materiality of applications, defined by the BOE survey, quantitative size-based measures, including exposure, book or market value, number of customers serviced or covered by the use case are notable. And in addition, there are qualitative factors vis-à-vis the purpose of the model and its relative importance to informing business decisions and considering the potential impact on the firm’s solvency and financial performance. Of the total number of use cases reported by the respondent firms, 62% were rated low materiality, 22% as medium, and 16% as high. Low and medium materiality use cases were most common in operations and information technology (IT), whereas high materiality use cases were common in general insurance, risk and compliance, and retail banking.
This data, when analysed in the Indian context, could reveal either a planned decision to adopt low material use cases with a view to demonstrate some early wins or that firms are focussing on other infrastructure blocks including building data foundations. What could be an adverse factor is that the low materiality use cases are more in the operations and IT areas.
Let us now examine the third important aspect of the framework, that of governance and accountability. The BOE survey covered the range of governance frameworks over a variety of approaches used by respondent firms. The most used framework, control or process specific to AI was to have an accountable person or persons with responsibility for the AI framework. This factor was closely followed by the second approach using an AI framework based on principles, guidelines or best practices and data governance. With regard to data management, the respondent firms believed that it was the key to governance. It is a major concern, however, that there is over-dependence on data science teams who are responsible for data ethics, bias reliability, and authenticity and fairness. In this approach, firms tend to use in-house databases which are for internal consumption rather than for customers and distribution partners.
The last key ingredient to sound governance for adoption of AI is the firm’s assessment of their own or third-party models. The aspects in the governance framework which are assessed include business need, evaluating how appropriate a particular type of model is to the business objectives. In the case of Indian BFSI firms use complexity tests, some of which are built into existing processes and some of which are AI-specific. AI-specific tests include consideration of methodology, data, complexity of code, interoperability, parameter count, and frequency of use. Complexity of data is also a central factor, particularly where large and multi-dimensional or multi-model data sets are involved. One very interesting aspect observed in firms worldwide is the understanding of AI technologies implemented in their operations. It is believed that a large number of firms have a partial understanding of the AI technologies used and a small number have near-complete understanding, underlying a major weakness which needs to be addressed over a period of time to strengthen governance and AI adoption.
In conclusion, to make AI adoption deeper and wider, a structured approach is an important factor. This needs a robust and dynamic framework to critically examine the materiality of the use case and a strategic approach to migrate from low materiality to high impact use cases and a governance framework making person/s accountable for framework as well as the use cases. Companies must address the aspect of partial understanding of the use cases or technologies used by way of creating awareness and gradually moving to making businesses own the responsibility for use cases.