In the last decade, Artificial Intelligence (AI) has emerged as the game-changing technology in the insurance sector. Apart from driving data transformation, it has been key in creating more efficient claims application and management systems as well as augmenting hyper-personal insurance products and services. But perhaps its most significant impact lies in risk management, particularly in claims and underwriting, where it is being utilised along with other technologies, such as Machine Learning (ML), to identify and minimise risks, detect frauds, and find a balance between risks and opportunities.
Optimising risk selection
Insurers are leveraging AI to identify underwriting risks and optimise risk selection. Smart algorithms comb through industry databases to cull pertinent data on customers, efficiently segregating them into pre-decided pricing categories. AI-based risk detection is utilised to identify credit risks, governance and compliance risks, operational risks, market risks, liquidity risk, trading risks, cyber risks, and the criminal risks, such as fraud or money laundering.
Embedded AI and real-time integration with industry databases have also helped to make the process of underwriting, including risk selection and pricing, faster and more efficient, significantly improving customer experience. For insurance companies, these technologies are fast emerging as a key competitive tool for customer acquisition and retention. Given the prominence of IoT and tracking devices in our life, and their access to precise and critical data, AI-related technologies will gain further importance in data analysis, risk selection, and pricing.
Smart claim processing
From chatbots for quick resolutions to ML applications, intelligent tools have completely overhauled the claims processing, making it more efficient while reducing risks. When it comes to risk management, data analytics has gone a long way in automating fraud detection, identifying patterns in claim volumes, and further strengthening loss analysis.
One of the biggest concerns for an insurance company is fraudulent claims. Investigating each claim can take up time and valuable resources. Today, visual analytics, involving the analysis of pictures and videos, has sped up the processes. Insurance companies can carry out preliminary investigations with minimal resources while relying on highly precise data, thereby weeding out fraudulent claims.
Predictive risk management is a crucial aspect of any insurance business. While underwriters perform due risk selection when deciding pricing, there is only so much data that a human can process. With the massive amounts of data at our disposal today, predictive analytics has necessarily been taken over by AI-based technologies. Smart predictive algorithms can scan through data to identify patterns in outlier claims- cases that result in unexpected huge losses.
This allows insurance companies to plan their policies so as to reduce chances of outlier claims. Predictive analytics can also help identify common risk factors to incentivise safe behaviour, thereby reducing overall claim volumes. For instance, health insurtech looks at hospitalisation data to identify high risk lifestyles. Consequently, the insurance company can incentivise safe practices that reduce the chances of hospitalisation among its customers.
One of the biggest challenges presented by AI-based solutions is in fixing liabilities. The shift from human to technology in decision making creates a grey area when it comes to decision making that could eventually lead to governance and compliance issues. As embedded AI technologies become a critical component of the underwriting process, we must be aware of unintended biases that can arise out of their implementation. While algorithms are touted as failsafe mechanisms to calculate risks, these must be applied keeping in mind certain social-cultural factors, and this is where machines can make mistakes.
Failure to account for these factors can give rise to two main liabilities – bias in claim settlement and discriminatory underwriting. Insurtech algorithms decide underwriting pricing based on factors like gender, creditworthiness, and social class. The model output may carry bias against any one factor even if the other variables meet the desired standard. Similarly, in the case of claims, it can dismiss meritorious claims based on an error in fraud detection.
Human-AI collaboration is not just important for ensuring the ‘human’ factor, it is also a necessary risk mitigation strategy. While machines can perform complex calculations, we need humans for emotional intelligence, such as to identify biases, and to ensure human-centric outcomes, creating a value add in both products and services.
An article published last year pointed out the vulnerability of AI-ML technologies to unintentional and intentional risks, calling for human oversight of critical decisions. For instance, human intervention is necessary to determine patterns of bias or discrimination in claims or underwriting. This also opens up the possibility of new business models addressing future liabilities that may arise out of AI usage, such as robo-advice.
AI and Blockchain
AI has also been instrumental in complementing blockchain transition in the insurance industry. The digital distributed ledger format of the Blockchain decentralises data, easing access, and ensuring transparency. AI automates Blockchain data gathering, allowing adjusters to quickly settle claims with lower risk of fraud. Similarly, IoT with blockchain can drive hyper-personalisation while reducing the risk of discriminatory underwriting by providing the underwriter with highly accurate and relevant data. In this manner, it can further leverage hyper-personalisation of insurance products.
More importantly, the transparent nature of the blockchain minimises risks of fraud, while optimising risk selection. For instance, blockchain augmented AI-powered healthcare analytics ensures accurate health record keeping, thereby helping reduce risks associated with underwriting or claims selection.
With the growing prominence of embedded AI-based technologies in our life, insurers have access to unprecedented amounts of data. This has provided them with unparalleled insight into their consumers, allowing them to improve customer experience, develop personalised products, and add value to the insurance value chain. However, we must stay vigilant towards the risks posed by the use of such technologies. It is vital that insurance companies undertake frequent internal risk evaluations to assess the safety of AI systems and prepare against any possible failures.
Authored by Hersh Shah, CEO, India Affiliate, Institute of Risk Management & Rohit Boda, Managing Director, J.B Boda Group