Darran Simons, FICO
Is there any area of modern life untouched by Artificial Intelligence? AI has certainly started to make a positive impact within insurance, and the industry is excited to see what the future might hold.
AI adoption within insurance has so far focused on risk management, advanced fraud detection beyond traditional modelling, and personalised policyholder experiences. Insurers are increasingly embracing AI’s potential for broader application, often combined with existing technologies and processes. For example, we expect to see more blended approaches in developing risk models, leveraging both AI and established modelling techniques. This will ensure explainability, which is crucial for regulatory compliance and building trust with policyholders.
While Generative AI has elevated discussions of AI, we’re still a long way from fully leveraging its power, or even understanding the full potential. Looking beyond Generative AI, it is exciting to see the ways advanced analytics and AI-powered systems are driving decisions and improving customer journeys. The insurance industry has made tremendous progress, and while insurers are a far cry from Netflix or Uber in their use of AI, it is powering a lot of digital transformation and legacy modernisation.
AI has the potential to drive hyper-personalised experiences, which are increasingly becoming the norm as customers expect to be treated as individuals. And it doesn’t just bring benefits to customers, it also helps with cost reduction and revenue growth. A Gartner survey has shown that brands risk losing more than 1 in 3 of their customers if personalisation is poor.
However, many organisations currently entrust their AI strategy to business owners or software engineers who may have little to no understanding of the intricate mathematics and risks associated with AI algorithms. While stringing together AI may seem straightforward, building responsible and safe AI is a far more challenging task. Corporations that define responsible AI model development standards will be able to use the appropriate types of AI for various applications, ensuring regulatory compliance and achieving optimal consumer outcomes. This corporate defined AI model development standard is crucial for creating and implementing AI models that are both effective and responsible.
Not all plain sailing
While AI is expanding into new roles within insurance, many businesses appear to be struggling to operationalise the technology. That is, moving the model out of the analytic laboratory in which it was conceived, and making it available as robust, tested, runtime software that accurately computes the model to inform a timely decision. Gartner also reported that only 10% of organisations have put generative AI solutions into production, though some sources put the number much lower.
This is largely an execution problem. Models take months to get into production, with the delays reducing competitiveness and model value. Advances in analytics, including AI and machine learning, are critical to make better decisions when margins are tight, as well as to improve customer retention and profitability. Utilising a platform that accelerates model deployment without requiring recoding can build responsible AI models efficiently without placing burden on IT.
Explainability is another big challenge. For AI models to get real traction with insurers, they need to be explainable. Legislation is coming as industry regulators get to grips with the need to protect consumers and businesses from the ‘computer says no’ aspect and ensuring AI is used responsibly. But it’s not just about legislation, it’s about understanding the way the model makes decisions so that insurers can be confident about applying it to more use cases the vast majority of explainable AI techniques are not sufficient, requiring critical use of interpretable AI.
The future of insurance AI
As insurers look to adopt responsible AI, there is another development that will be crucial to success. Enhanced collaboration between business stakeholders and a data science team ensures that the right solution addresses the correct problem.
Closer interaction and collaboration among data scientists, business users, and IT could also lead to superior models and quicker execution. For example, FICO Platform serves as a collaborative environment where such groups can work together seamlessly. Our clients report that using the same system to build analytics and decision strategies and simulate their results enables faster adoption of both new models and strategies based on them.
Building for long-term pay-off
Insurers that enable their business operations to be transparent and controllable are the ones that can truly capitalise on their AI investments. By fostering robust collaboration between business stakeholders and data science teams, leveraging platforms for seamless cooperation, and ensuring a deep understanding of both the complexities and responsibilities of AI, insurers can build superior models and achieve faster execution. In doing so, they can stay ahead of the competition and fully realise the transformative potential of their AI strategies.