During its annual conference on 27TH February 2024, the Association of British Insurers (ABI) launched its "AI Guide: Practical ideas for getting started with responsible AI" (the "Guide") for those working in the insurance and long-term savings industry in response to the developing regulatory framework. The Guide intends to assist insurers who are grappling with the issues raised by the use of AI by providing a starting point for engagement which includes key questions, known risks and good practice suggestions.
AI has the potential for considerable application in the insurance sector through the product lifecycle. The guide emphasises how AI can be used in that lifecycle. Examples include:
- Product design – the use of AI to identify patterns within underwriting portfolios, claims patterns and the opportunity to rebalance exposure
- Pricing and underwriting – personalised pricing based on individual risk profiles including driving behaviour and health, predictive underwriting offering tailored policies and augmenting actuarial models for mortality and morbidity
- Claims and benefits management – the use of AI algorithms to detect unusual patterns and behaviours linked to fraud, the automation of claims and predictive models to predict and assess the impact of natural disasters.
The Guide aligns with the UK policy paper on AI, defining the term by reference to the characteristics of adaptivity and autonomy. For those insurers operating within the EU, the classification system set out within the AI Act is also referenced.
The Guide sets out the five principles underpinning the UK's AI policy and uses them as the reference point for understanding AI-associated questions and risks. The Guide usefully sets out key questions that firms may wish to consider in terms of how AI will affect their business operations, followed by examples of good practice. However, the Guide also makes clear that the questions are "not exhaustive, and firms will need to consider how to apply the principles within their own business models and risk appetites."
The Guide also provides an AI risk taxonomy within the context of each of the AI principles and sign posts to existing regulations including Consumer Duty, FCA requirements, data protection and the Equality Act 2010.
