This piece is by Pete Allchorne and Jo Folan of the Strategic Advisory team at DAC Beachcroft’s claims division, CSG.
The claims process is an ideal environment in which processes can be automated. Whilst there is a need for human judgement at various stages of the claims process, some of the tasks that must be carried out are both formulaic and repetitive, and capable of being automated in order to drive speed and efficiency for the benefit of the customer.
Generative and Agentive AI have put an entirely new spin on automation. These are systems that can learn and improve over time and are potential game-changers for the claims process.
Those who embrace AI within the claims process will have a competitive advantage if used in the right way. Insurers are there to help people and businesses at a moment of jeopardy. The ability to demonstrate empathy, to apply human intuition and judgement, and to explain the rationale in making decisions that materially impact the outcome of the claim will remain vital ingredients to achieving the best customer outcomes – something that data driven decisions cannot replicate.
In order for human and machine to work together in harmony, it will be important to educate colleagues and involve them in the AI strategy. Training is essential to enable colleagues to understand the benefits of using technology to assist and empower them in carrying out their work in a more efficient and satisfying way. Understanding what AI is helpful for and how it can best be prompted and used to ensure that colleagues are supported in their role, whilst maintaining oversight of decision making will be critical. Ultimately, by delegating administrative tasks to an AI assistant, claims handlers will be more available to customers, and will be able to access the information that they need to take decisions more quickly and effectively.
Where this balance is not appropriately maintained, such that there is an over-reliance on automated processes, insurers may find themselves facing heightened risks. Insurers operate within a highly regulated environment and must ensure that they treat all customers fairly in the processing of claims, to include adherence to the FCA’s Consumer Duty requirements where applicable. Failure to ensure that humans are adequately involved in taking or overseeing key decisions risks insurers not having effective insight into whether they are achieving fair and lawful outcomes. Should any insurer find themselves in regulatory breach as a result of over-reliance on AI, this could cause significant reputational harm to their business, particularly if the breach was a systemic one.
It is all about achieving the right balance between human and machine, and that balance will be different depending on the part of the claims process in question. For example, AI can be used to quickly and effectively flag fraud indicators, but where there are potential fraud allegations to be made, these are nuanced decisions that will inevitably require the judgment, experience and intuition of an experienced claims handler.
Ultimately, the buck stops with insurers to act in good faith towards their customers, and therefore the final responsibility must remain with claims handlers to evidence that the right decisions are being made, based on experience, judgement and intuition. AI can be used to streamline processes and allow claims handlers to work more effectively but, in the event of an unfair or unlawful outcome for their customers, seeking to blame a computer algorithm would ring very hollow. The same would be true were a customer to complain about a poor outcome where decisions were taken by a machine with no human oversight or validation. An insurer would struggle to satisfy the Ombudsman about the fairness of any decision over which the insurer had no oversight and which it could not justify objectively.
Balancing efficiency and effectiveness will be a fine line to tread, and will undoubtedly change over time as the processing power of computers increases and AI evolves and becomes ever-more sophisticated.
Be the first to comment