AI in Claims; the good, the unknown, and the opportunity
22 November 2023
18 December 2023
Lee Elliston, Chief Operating Officer, Global Claims, Aspen Insurance Group
In an era where the insurance industry strives for innovation through technology, focusing more on data than ever before, the significance of talent and skills development cannot be overstated. This article explores how the convergence of technology, data and skills can shape a successful claims proposition. It introduces the concept and role of Intelligent Assistants (the new IA) as a new lens through which to view the use of technology in claims processing.
Redefining AI as IA in Claims
Rather than the often and more recently discussed hot topic of Artificial Intelligence (AI) in claims, we propose a shift in mindset to Intelligent Assistants (IA). This perspective emphasizes the practical application of AI within the claim’s lifecycle, promoting a balanced approach that maximizes customer value, interaction and product promise, without excluding the human factor.
Careful Integration of AI into Claims Models
To fully realize the potential of AI, it is crucial to embed it thoughtfully within the structure of insurance operations. AI should be integrated seamlessly into the claims serving model, strategically applied during critical data touchpoints in customer and product decision-making processes. This approach establishes a partnership between data, algorithms, claim specialists and insured parties, creating business value and enhancing customer experiences.
The Good: Advantages of AI in Claims
Integrating IA and merging with AI algorithms and machine learning can advance data skills and trust, offering such benefits as:
- Swift triaging of claims to the right specialists at the point of First Notice of Loss (FNOL)
- Recommending data-driven claims coverage based on risk and policy data
- Using historical data for quicker claim settlements and more precise reserving
- Improving fraud detection and alerts during the claim lifecycle
IA enables a shift to semi-supervised AI, empowering claims specialists to be more effective in their oversight of a portfolio. They can intervene and review the intelligence provided to add value to the decision and outcome. This semi-supervised approach allows automated activity and decisions to be handled at FNOL and post-coverage determination, transforming the claims specialist's role away from that of a transactional processor. This shift reduces the blurriness between value-added activities and processing, enhancing effectiveness and efficiency within a claims team. It also advances the portfolio role in collaboration with underwriting colleagues, freeing up time for higher-skilled and more valued activities.
The Unknown: Challenges and Risks
Implementing AI is not without risks and challenges, including:
- Liability concerns: with the deployed AI or IA potentially liable for decisions or outcomes, there are questions about where the liability rests in interpreting data and making decisions.
- Declared and discoverable requirements: with forensic analysis required to uncover causes, liabilities, and harm, this is an uncharted path with implications for legal costs and outcomes.
- Data and algorithm security issues: with the risk of breaches during a data exchange with third parties, this potentially compromises personal data and damages trust in the insurer's security.
- Potential claims leakage: when AI is used in an immature data model or where claim specialists lack data confidence, there is the threat of potential claims leaking.
- Immature, inaccurate and unmanaged data model: with AI's trustworthiness dependent on strong data foundations, including high data quality, well-structured data and well-managed data sources.
The development of generative AI holds immense potential, enabling a quicker transition from proof of concept to scaling up. However, uncertainties surround the practicality of generative AI solutions, despite their transformative potential. A cautious approach to generative AI that, forming part of a data evolution strategy, allows insurers to navigate this uncharted territory. Small concepts, established by each function in an insurer, enable testing and learning from practical applications throughout the insurance lifecycle, generating a trusted and road-tested benefit case for scaling up across functions, products, or business lines.
This partnership of AI with specialists creates the role of the new Intelligent Assistants, competing with the Independent Adjusters. Today’s IA can provide data insights, recommendations, human touch processing and engagement activity, without fully replacing human decision-making or the ethical and empathetic fairness, validation, or negotiation. AI is a powerful tool that becomes more influential when influenced by human language, skills, and knowledge. It can enhance and change behaviors, generating positive use cases and outcomes within the claim’s role and other aspects of the insurance lifecycle.
The regulation, application, and governance of AI, particularly within Lloyd’s, presents unique challenges. For example, the adoption of AI by one Lloyd’s insurer can impact and bind many others on risk. The fairness and consistency of AI's use requires careful consideration. Collaboration between such organizations as CII, Lloyd’s and LMA is recommended to establish a business, market, and consumer lens for the application of AI within claims. This collaboration can identify opportunities to establish policies, and support the development and upskilling of specialists, thereby ensuring a fair and consistent implementation of AI can be achieved without restricting innovation.
At Aspen, the following are several ways the business is preparing, both in partnership and to differentiate our value:
- We are investing in and delivering our Single Version of the Truth data architecture to replace legacy data warehouses, with robust and dynamic master data management and governance.
- We are establishing use cases and key factors to help remove bias.
- We are running a proof of concept on AI and algorithmic claims adjustment and intelligence, using claim, customer, and 3rd party data to deliver data truth to support the presentation and quantify the impact as an instant claim assessment.
- We are delivering our data first strategy, with data models being key foundations to further road test how we leverage and embed our data for global use in 2023 / 24.
In conclusion, AI is here to stay and to be leveraged. And the new IA is to be born and develop, with insurers and claims management companies evolving their operating model and workforce strategy, particularly where their data first strategy and foundations enable it. Further, the judicious use of Intelligent Assistants – combined with human expertise – can help to revolutionize claims processing and the insurance industry. As the industry embraces this transformation, the key lies in collectively shaping a future where AI contributes to the new Intelligent Assistants, to enhance the overall customer experience and with it build trust from a source of data truth that is shared between client, broker, 3rd party, and (re)insurer.