The Insurance Regulatory and Development Authority of India has constituted a seven-member working group to examine the adoption, governance and oversight of artificial intelligence in the insurance industry. It has reportedly been given three months to submit its recommendations.
The initiative is timely. Artificial intelligence is already influencing underwriting, claims, customer service, fraud detection and distribution. However, the insurance sector must avoid treating AI merely as another technology project. Its real value will be measured by whether it improves protection, reduces delays, makes products more affordable and strengthens public confidence.
Underwriting is one of the most promising areas for adoption. AI can analyse medical information, past claims, property characteristics, weather patterns, driving behaviour and industrial risk data more efficiently than conventional systems. It can help insurers identify risk accurately and offer coverage that is better aligned with the customer’s circumstances. For small businesses and individuals who are often excluded because insurers lack sufficient information, responsible use of alternative data could improve access to insurance.
Claims offer an even greater opportunity. AI can read policy documents, classify claim papers, identify missing information and prioritise straightforward claims for faster settlement. In motor insurance, image analysis can assist in assessing visible vehicle damage. In health insurance, automated scrutiny can help detect inconsistencies between hospital bills, treatment records and policy terms. Human claims officers should then concentrate on complex cases requiring judgement and empathy.
Fraud control can also improve considerably. AI systems can identify unusual claim patterns, repeated use of documents, suspicious provider networks and connections between seemingly unrelated transactions. This can reduce leakage and ultimately help control premiums. Nevertheless, an algorithmic alert must only trigger further investigation. It should not become an automatic justification for rejecting a genuine claim.
Artificial intelligence can also strengthen customer service. Multilingual virtual assistants could explain products and claim procedures in simple regional languages, provide policy information and guide customers in submitting documents. This would be particularly useful in smaller towns where access to trained insurance personnel remains limited. However, customers must always have the right to speak to a human representative.
AI can support intermediaries rather than replace them. Agents, brokers, surveyors and other professionals could use AI tools to compare products, identify protection gaps, prepare risk reports, verify documents and provide quicker service. Such tools should enhance professional judgement and productivity, not reduce insurance advice to an automated sales script.
The risks, however, are equally significant. AI can reproduce bias contained in historical data, unfairly discriminate against certain customers, generate inaccurate information or make decisions that cannot be explained. Excessive reliance on automated underwriting could result in arbitrary premium increases or denial of coverage. Similarly, automated claims systems may reject legitimate claims because the case does not fit an established pattern.
IRDAI should prescribe clear accountability. Every insurer using AI must have a board-approved policy, documented models, periodic bias testing, cybersecurity controls and pre-deployment and post-deployment audits. Customers should be informed when an important decision has been substantially influenced by AI and must have a simple mechanism to seek human review. Insurers must remain responsible even when technology is supplied by an external vendor.
Authored By
Dr. Rakesh Agarwal
Editor
The Insurance Times
