Artificial intelligence is changing the way banks assess credit, detect fraud, manage investments and interact with customers. Its growing integration with decision analytics is moving financial institutions beyond conventional reporting and forecasting towards predictive and increasingly autonomous decision-making. The potential gains in efficiency, accuracy and productivity are considerable, but so are the challenges of ensuring that the decisions generated by these systems remain explainable, reliable and accountable.

AI's relationship with decision analytics is analogous to that between the brain and the rest of the body. Decision analytics combines data, statistical modelling and management science to evaluate alternatives and choose the best course of action under uncertainty. AI provides the computational speed and pattern-recognition capabilities that allow these processes to operate on much larger datasets and increasingly in real time.

Traditional analytics often stops at describing what happened or forecasting what might happen using periodic information. Machine-learning models can identify complex patterns and hidden relationships that human analysts or conventional software may miss. In banking and finance, this is enabling a significant shift towards predictive and autonomous choices.

Smarter Decisions
Credit assessment is among the most important applications. AI-driven analytics can process alternative datasets, including utility bills, rental histories, cash-flow information and digital footprints. Advanced machine-learning models can identify non-linear risk factors that conventional credit-assessment methods may overlook.

Studies indicate that banks integrating these systems can increase loan approval rates by as much as 78% while reducing default risks by 20%. Automated processing can compress loan approvals from days to minutes.

Fraud detection and anti-money laundering offer another example. Traditional rule-based systems have generated false-positive rates of 90–95%, imposing substantial costs on compliance teams. AI systems combining supervised learning and anomaly detection can evaluate hundreds of data points simultaneously, including transaction velocity, internet protocol addresses and device fingerprints.

Such systems can reduce false positives by 40–70%, allowing compliance teams to focus on genuine threats rather than sorting through large volumes of unnecessary alerts.

Investment management is also changing. AI models continuously reassess asset correlations, while natural language processing examines earnings-call transcripts, news sentiment and regulatory filings. Portfolio decisions can incorporate real-time volatility, tax implications and clients' risk preferences.

Yet, these capabilities introduce a fundamental difficulty. Deep-learning models are often black boxes. They may produce accurate decisions without explaining how those decisions were reached. When a loan is rejected or an account flagged as fraudulent, financial institutions must be able to provide an auditable explanation.

Explainable AI frameworks can generate reason codes that help compliance teams understand model decisions. Their development is important because greater computational sophistication cannot substitute for accountability.

Indian Experience
India is at the forefront of the digital revolution, and AI-driven decision analytics is becoming an important part of banking operations.

Banks are increasingly using alternative information, including utility bills, GST filings and cash flows, to assess micro, small and medium enterprises and borrowers with limited conventional credit histories. Machine-learning models monitor transactions for anomalies, while chatbots and virtual assistants provide round-the-clock customer services.

Surveys of Indian banks show that all have implemented mobile and internet banking with AI features. Around 75% offer online account opening, digital know-your-customer services and digitally enabled doorstep banking. Digital lending is available at 60%, payment aggregator services at 50%, and AI- or machine-learning-based chatbots and virtual assistants at 41%.

Private sector banks are leading technology adoption, although public sector banks have also made progress. Mergers among public sector banks appear to have encouraged the adoption of chatbot technologies already used by acquiring institutions.

The benefits are becoming measurable. An AI-assisted review of Indian banks' annual reports reveals examples of monthly savings of 14,500 person-days, reductions of 25–30% in customer acquisition costs, a 40% decline in branch waiting times and a 50% reduction in compliance-monitoring time. Account opening has also been shortened to less than a day in some instances.

These are gains from AI and other digital technologies, rather than AI alone. Nevertheless, they demonstrate how technological adoption can improve operational efficiency and customer experience.

The expansion of model-based lending through collaboration between banks, non-bank financial companies and fintech firms also demands caution. Financial institutions should not rely solely on preset algorithms. Models must be robust, and their resilience should be tested periodically.

The best results are obtained when AI and human skills complement each other.

Supervisory Challenges
Central banks are themselves adopting emerging technologies to strengthen monetary policy, financial supervision and regulatory compliance.

The Reserve Bank of India has been using big data analytics, AI and machine learning across monetary policy, research and data management. Hybrid models combine traditional statistical methods with machine-learning techniques for forecasting and nowcasting. Online food prices help track inflation, remote-sensing data assist assessments of crop production, and media sentiment is analysed to evaluate the effectiveness of central-bank communication.

On the supervisory side, the RBI's Advanced Supervisory Analytics Group applies machine-learning models to social-media analytics, know-your-customer compliance and assessments of governance effectiveness. DAKSH, its advanced off-site supervisory monitoring system, is helping digitalise supervisory processes.

These developments are making central banking more data-intensive and expanding the scope of regulatory oversight. They also require greater attention to data quality, governance and the reliability of analytical models.

Cybersecurity presents a particularly important challenge. Existing requirements provide for cybersecurity operations centres at scheduled commercial banks and certain urban cooperative banks. However, institution-specific surveillance may overlook threats that affect the financial sector as a whole.

A sectoral security operations centre is envisaged to correlate incidents and information from multiple institutions and generate actionable intelligence. Centralised monitoring could help identify common vulnerabilities and emerging attack patterns.

Third-party concentration is another concern. When numerous financial institutions depend on a small number of technology providers, operational failures or cyberattacks can create systemic vulnerabilities.

An exercise mapping IT service providers used by scheduled commercial banks has informed a proposal for joint vendor audits under the Indian Banks' Association. Such audits would enable more comprehensive and standardised assessments of technology providers.

Cloud computing and software-as-a-service arrangements introduce further questions concerning data security, privacy, operational continuity, interoperability and auditability. Financial institutions must recognise that technology risks can extend beyond their own organisational boundaries.

Human Accountability
The regulatory challenge is to encourage AI-driven innovation while safeguarding financial stability and consumer trust.

The RBI's Framework for Responsible and Ethical Enablement of Artificial Intelligence, introduced in 2025, seeks to establish this balance through seven guiding principles and six operational pillars.

Trust forms the foundation. AI systems must be reliable, transparent and secure. Human judgement and citizen welfare must take precedence over automation, while the institutions deploying AI remain accountable for the decisions made through these systems.

Fairness requires that outcomes do not discriminate against users. Understandability demands that decision-making logic can be explained to customers and regulators. Safety, resilience and sustainability require continuing protection against cyber and operational risks.

Importantly, the framework also emphasises innovation over restraint. The objective is not to impose unnecessary barriers on technological development, but to enable responsible adoption.

Its operational pillars cover infrastructure, policy, capacity, governance, protection and assurance. These encompass secure data systems, clear deployment and risk-management policies, workforce training, board-level oversight, consumer protection, independent audits and regular testing.

Training data must be extensive, accurate and diverse to minimise bias, while algorithms must remain auditable. Institutions also need to invest in reskilling and upskilling employees as technology changes the nature of financial work.

The responsibilities of regulators are expanding alongside those of financial institutions. Oversight authorities must strengthen surveillance capabilities, update regulatory frameworks, engage stakeholders and improve consumer education.

AI is redefining policymaking, re-engineering work processes and creating new capacities to meet demands for greater speed, convenience and affordability. Fintech and bigtech providers are introducing products and services that can transform the financial system, but their innovations also create risks that must be understood and managed.

The task is to harness these technologies without compromising security, privacy, legal compliance or ethical standards. Human judgement remains essential, not as an obstacle to automation, but as a necessary complement to it.

Ultimately, regulation can either stifle or foster innovation through the rules and standards it establishes for cooperation, interoperability and competition. Getting that balance right will determine how effectively banking and finance benefit from the AI revolution.

(Based on a lecture delivered at the National Institute of Bank Management titled “Artificial Intelligence and Decision Analytics for the Banking and Finance Sector”. This is the is concluding part. Part 1 dealt with the economics of AI.)