Artificial intelligence and risk management in the insurance sector
August 2021 | TALKINGPOINT | RISK MANAGEMENT
Financier Worldwide Magazine
August 2021 Issue
FW discusses artificial intelligence and risk management in the insurance sector with Marlene Dailey and David Mamane at RSM Consulting LLP.
FW: How would you describe the adoption of artificial intelligence (AI) by the insurance sector in recent years? Is AI technology now a prevalent, integral part of the risk management process, or do you believe we are still in the relatively early stages of adoption?
Dailey: AI adoption is accelerating rapidly among insurers. The pandemic has forced insurers to shift from being heavily reliant on in-person interactions, appraisals and inspections. We are still in the relatively early stages of adoption. While insurance executives understand that AI is reshaping the competitive marketplace, there are still significant roadblocks that many insurers face in optimising the huge amounts of data within their legacy systems to realise the full value of AI technology.
Mamane: Over the last decade, insurance companies have been increasing their focus on data transformation initiatives, all in the interest of building a robust data foundation. The richness of data collected by insurance companies goes beyond policy and claims data and has been further enhanced thanks to the rise of telematics, usage-based insurance products and internet of things (IoT) devices. To operationalise this data, companies are beginning to deploy modern, self-service analytics platforms to empower business users. This evolving data ecosystem creates a perfect breeding ground for innovation in AI and machine learning (ML) across many operational processes of insurance companies, including underwriting, pricing, claims and more. Insurance companies that tap into this potential can set themselves up for future success. Property & casualty (P&C) insurance carriers, for instance, called on their data assets early in the pandemic as policyholders reassessed their insurance needs, decreasing vehicle usage and increasing residential coverages. From a distribution channel perspective, more consumers shifted to using digital platforms and growth in digital adoption of online portals and mobile application created new potential touchpoints between carriers and their customers, as well as a wealth of structured and unstructured data that is vital to the adoption of AI in insurance.
FW: In what specific ways is AI being used to augment insurance companies’ risk management activities?
Dailey: AI is transforming risk management, particularly in the areas of claims and underwriting. AI is being used to help insurers with risk management tasks like recognising underwriting risks and detecting fraud more effectively. Some insurers, for example, are leveraging AI’s natural language processing and advanced analytics capabilities to extract pertinent risk information from emails to identify underwriting risks and optimise risk selection. In claims, insurers are leveraging AI technology such as ML and advanced analytics to predict potentially fraudulent claims, elevating the adjuster’s capabilities to conduct thorough investigations.
Mamane: AI and ML techniques can also have profound applications in corporate functions of insurance companies such as finance, risk management and actuarial. As with many other critical business functions, corporate functions are continuously looking to technology to enhance the efficiency and accuracy of their processes. Finance departments can leverage AI-enabled tools like intelligent automation to perform tasks like order processing, journal entries and reconciliations. Risk management departments can use ML techniques to move away from monitoring lagging performance metrics to uncovering forward-looking key risk indicators. Finally, actuarial departments continue to make significant advancements in customer segmentation, pricing and reserving models with ML techniques.
“As AI technologies become embedded in the underwriting process, both from a risk selection and pricing standpoint, it is critical for insurance companies to monitor process results for unintended biases.”
FW: To what extent can AI reduce costs and improve decision making and productivity around risk management processes?
Dailey: The power is in prediction. As AI has matured, the technology has become capable of delivering meaningful predictions by assessing and analysing forms of structured and unstructured data to identify patterns, processes and anomalies therein. AI applies science to minimise the risk manager’s assumptions and behaviours and develop better predictions. Risk managers can use this predictive power to improve decision making, boost productivity and reduce the frequency and severity of allocated loss adjusting expenses (ALAE) in claims.
Mamane: AI can reduce costs significantly in the insurance industry, but cost reduction is not often the primary focus of an AI strategy. Underwriting is an often-promoted area where AI and ML techniques are used to improve risk selection and overall customer experience when purchasing a policy. Many insurance companies have invested significantly in research and development (R&D) and advanced analytics capabilities with a primary focus on outdoing their competitors on customer acquisition and retention. However, AI applications focused on underwriting do not reduce costs in aggregate at the industry level, but rather distribute the same total costs across all market participants. Insurance companies that invest in AI technologies focused on risk control and mitigation can achieve a reduction in insurance claims and create a sustainable, long-term competitive advantage.
FW: When implementing AI into existing systems, what typical challenges might companies in the insurance sector expect to confront? What steps can they take to overcome these challenges?
Dailey: The typical challenges we see within the insurance sector start with planning and strategy. These transformations are typically IT-led, with a focus on AI technology solutions, such as design apps, dev ops, data migration, and all things tech. The business is involved when it comes to functional requirements, but business leaders are usually thinking about how to retrofit the current capabilities of the business with the AI solution. It is important to think of AI itself as not just a specific solution, but a tool. Employee enablement – who is going to do the work – and providing value to customers can get lost in the shuffle during implementation if business leaders are not intentional. Planning and strategy is key. Know your people, your customers and your data. This is easier said than done, but there are steps insurers can take to make implementation of AI solutions smoother. Transformation starts at the top, so make sure your strategy aligns to your vision. Communicate early and often to cultivate a culture of understanding about what the future looks like. Know where you are and the pace at which you need to move to ready your organisation. Use an independent resource to conduct an initial readiness assessment across the organisation. This helps to eliminate bias and groupthink. Lastly and most importantly, take care of your people so that they can take care of the mission.
Mamane: Access to quality data is also a key challenge faced by the insurance industry in the implementation of AI into existing systems. Most insurance companies still operate today with one or more legacy IT systems, often from prior M&A activity or simply due to the large investment required for system modernisation. The reality is that for many insurance companies, even some of the largest carriers in the world, it can feel grim, and it is often very challenging to know where to start when implementing AI. Overcoming the data quality challenge is not a trivial exercise for insurance companies. Historically, companies have made significant investments in data foundation and transformation initiatives to extract data from legacy IT systems and transform it into usable data assets. However, converting legacy systems to harness historical data should not be an impediment to implementing an AI strategy. Legacy systems often lack the metadata present in modern insurance platforms, which can be valuable in AI applications and often provide more predictive power than traditional data elements. When implementing AI into business processes, insurance companies should focus on developing solutions for processes found in modern insurance technology stacks, where data recency is critical and legacy data is less relevant. Do not let your legacy issues define your AI strategy.
“Transformation starts at the top, so make sure your strategy aligns to your vision.”
FW: What regulatory considerations do insurance companies need to make when utilising AI technology? How might compliance issues affect AI deployment strategies?
Dailey: Insurers should keep up with the introduction of new regulatory changes and frameworks. Regulators globally realise that AI is changing the insurance landscape, and as a result these bodies are developing data privacy and consumer protection frameworks such as the General Data Protection Regulation (GDPR), changing the ways in which companies process and analyse personal information on their customers. Compliance is required whether a violation is intentional or unintentional. In the US, the National Association of Insurance Commissioners (NAIC) – which provides guidance and sets standards for insurance state regulators – developed insurance-specific AI guiding principles known as FACT, as insurance companies must be fair, accountable, compliant and transparent. Operating within a unified regulatory and cyber security framework is becoming increasingly important as data becomes more widespread and exchanged across ecosystems. Insurance companies must demonstrate their ability to ensure confidentiality and avoid outcomes that are either unfairly discriminating or otherwise violate legal requirements, such as privacy and data security laws and regulations.
Mamane: Although often based in similar principles, regulation in the insurance industry can vary widely country by country and, in some cases like the US, state by state. When designing an AI strategy, insurance companies must have a holistic understanding of the regulatory environments in which they operate and ensure compliance with the various guidelines regulators may impose related to personal information. For example, in many jurisdictions, regulators continue to forbid the use of customer creditworthiness, gender and race information in insurance pricing models. AI applications should implement similar restrictions to avoid undue regulatory scrutiny. When it comes to demonstrating compliance with privacy and data security laws and regulations within the context of an AI strategy, failure can result in significant financial, reputational and strategic risk to an insurance company. Regulatory fines are often the first consequence of noncompliance and fines alone can cause significant setbacks from a return on investment (ROI) standpoint for any AI initiative. But perhaps the biggest ramification of failing to comply with privacy and data security laws and regulations is the damage to an insurance company’s reputation, which can impact the long-term sustainability of the business. Compliance needs to be at the forefront of AI deployment strategies to avoid the pitfalls that can cause irreparable damage to an insurance company’s brand.
FW: Going forward, as responsibility increasingly shifts from humans to machines, what liability scenarios may need to be addressed?
Dailey: Many organisations realise that providing AI solutions that create ethical, technological or regulatory issues may jeopardise their brand. As more insurers implement AI technology solutions in their decision-making processes, bias and prejudice may become more prevalent. Leveraging software or surveillance capabilities that identify ML algorithm biases can help but may not entirely solve the problem. At the end of the day, AI needs human intelligence. Human oversight is needed at critical business decision points.
Mamane: Two main liability scenarios in the insurance value chain come to mind as we increasingly shift to machine-run processes – discriminatory underwriting actions and biases in claim settlement. As AI technologies become embedded in the underwriting process, both from a risk selection and pricing standpoint, it is critical for insurance companies to monitor process results for unintended biases. Say, for example, a machine-learning algorithm is used to set the price for a given risk. The model input data may fully comply with various regulatory guidelines and restrictions around the use of characteristics like race, gender and creditworthiness. However, the model output may still produce a bias related to one of these characteristics if combinations of other variables are a good proxy for the restricted variable, leading to various compliance-related issues. The second liability scenario that needs to be addressed involves AI technologies in the claims process. Fraud detection models in claims are often touted as one of the leading use cases for AI in insurance. However, models may present biases in fraud detection and, without appropriate human intervention, customers could have their meritorious claim denied by a machine. In both liability scenarios, a robust AI monitoring programme can help to continuously test for unintended model biases.
FW: What essential advice would you offer to the insurance sector on using AI capabilities to better manage the risks they face?
Dailey: The speed at which an insurance company transforms depends on the organisation’s readiness, so companies should ‘PACE’ themselves – plan, assess, communicate and educate. Plan and build a strategy. Assess your organisation’s readiness, data and technology. Communicate early and often to align the organisation. Educate and train your people continuously.
Mamane: For larger insurance companies that can afford to invest in AI talent and technologies, do not try to solve every problem with AI. Focus your attention on ‘low-hanging fruit’ use cases that can generate return without undue risk or disruption to the business. For smaller insurance companies that have yet to invest in AI, it is not too late to start. AI technologies have come a long way in the last decade. Advancements in enterprise analytics platforms that leverage automated machine learning can equip business users with the tools to design AI solutions without the need for in-house data scientists.
FW: What are your predictions for AI as a risk management tool for the insurance sector? What innovations and applications do you expect to see in the years ahead?
Dailey: We expect more insurers to use cognitive analytics to access unstructured data to reduce biases in decision making and enable greater visibility into an organisation’s true risk exposure. I think we will see a fundamental change in the role of risk management, where humans and machines collaboratively work together. Oversight and risk management will be built into systems and continuously monitored by virtual risk agents, while humans will continuously monitor the systems and algorithms.
Mamane: We also anticipate AI applications and intelligent assistants will become commonplace across the insurance company’s technology stack, enabling insurance professionals to make more informed decisions in managing risk across the business. Insights from AI technologies will also drive a better understanding of the customer and their needs, allowing for a personalised customer experience throughout the insurance value chain. In the years ahead, I expect insurance companies will continue to invest in their data assets and expand the universe of data that they use to develop AI applications. In some cases, this expansion may happen through the acquisition of third-party data, but in many cases, it may be through partnerships and embedded insurance offerings with companies outside the insurance sector, like auto manufacturers and travel companies. These new distribution channels have the potential to unlock new insights into customers previously unknown to insurers and may uncover novel AI use cases that were not previously possible.
Marlene Dailey provides a range of property & casualty claims, policy administration and insurance management consulting services for insurers and reinsurers for all lines of business. She focuses on technology and digital services to develop innovative insurance services. She has led multidisciplinary teams in managing and delivering complex digital transformation programmes and brings her clients a matchless perspective having spent more than 20 years in the insurance industry in claims and underwriting operations. She can be contacted on +1 (919) 793 8521 or by email: marlene.dailey@rsmus.com.
David Mamane is a consulting director in RSM’s actuarial consulting practice and provides a diverse range of property & casualty (P&C) actuarial, insurance management and enterprise risk management advisory services to insurance companies and other large financial institutions. He has over 12 years of experience in P&C insurance pricing and reserving, economic capital and risk modelling, enterprise risk management, stress testing and data analytics. He can be contacted on +1 (647) 730 1325 or by email: david.mamane@rsmcanada.com.
© Financier Worldwide