Artificial Intelligence (AI) has never been more economical or more accessible, there is no longer any excuse for not adopting it — one way or the other. The insurance industry is well-known for its centuries-long resistance to change and lack of innovation, pinpointed by the continued use of face-to-face trading and rubber stamps in the Lloyd’s of London market. Despite always having been a data-intensive environment, insurance has been slow to adopt big data, let alone leverage it for business benefit. However, like it or not, the digital revolution is fast approaching.
In this blog we discuss the main impacts we think AI will have on the insurance industry — ranging from quiet evolution to eye-catching disruption.
Business processes: More efficient and customer-friendly
Akin to putting plasters on a dying dinosaur in the hope that it will make the animal feel better, insurers have started experimenting with AI in their existing day-to-day operations with focus on the core processes of distribution and claims. Menial, repetitive tasks are being digitized, but for the time being AI application is generally limited to augmenting humans.
Currently, the distribution chain between insured and carrier can have as many as 6–7 middlemen, including retail brokers, local insurers, MGAs, wholesale brokers and global reinsurers. Customer information is collected and examined in each step, leading to a great deal of human error and time-consuming manual work. By using AI techniques, manual entry and re-entry as well as the number of errors it leads to can be reduced. As the saying goes — algorithms never get bored the way people do.
Similarly, claims processes have stayed the same for decades with paper-based forms and files getting stuck as employees go off on holiday. Algorithms have the power to automate claims to a large extent, using technology to report a claim, capture damage, apply decision rules and communicate outcomes with the customer. In its “Insurance 2030” report, McKinsey & Company estimate that AI can reduce the current claims-related headcount in insurance companies by 70–90%.
Underwriting: Making risk evaluation and selection smarter
The heart of insurance, risk evaluation and selection, has come a long way since the times when each individual risk had to be reviewed and underwritten independently. However, underwriting in many places is still revered as an art, not a science, with decision-making mainly based on the personal judgment of underwriters and limited data collected from manual multi-page application forms. AI techniques and the addition of external data sources — e g satellites and IoT devices — will help underwriters determine which submissions to review to begin with (rather than all of them as is currently the case) in addition to giving a more holistic view of any risk, aiding profitable risk selection while automating large parts of it in accordance with pre-set decision rules.
Over time, we believe that the amount of submissions that require some form of human touch will fall significantly, most likely by 80–90%. In the future, Chief Underwriting Officers will be asking “Which data scientists did you have lunch with?”, rather than “Which brokers did you have lunch with?”.
Product development: Filling the current risk transfer gap
Insurance will only start moving from a declining dinosaur to a modern mammal for real when AI is applied to enable new product development in the industry. At present, corporate leaders in the developed world complain that effective risk transfer mechanisms are missing for the most important issues that keep them awake at night, such as reputational risk and information security risk. Similarly, developing countries remain underinsured to natural catastrophe risk due to a lack of models and data to support the models. Thirdly, SMEs and gig workers are a generally underserved segment of the market due to the inability of many insurers to accurately price their risks in a cost-efficient manner.
Applying AI to existing and alternative data sources will enable insurers to analyse risks they haven’t been able to understand previously — filling the risk transfer gap and helping customers mitigate the exposures that truly matter to them.
Value chain transformation
As a result of AI and additional data, we predict that the modern mammal of insurance will likely look radically different from its current dinosaur ancestor.
The transactional influence of insurance brokers is likely to shrink as their role in collecting and processing data is reduced. To compensate, brokers are likely to take on an even stronger advisory role going forward — banking on the large amounts of data they gather — as they battle to maintain ownership of the client relationship.
The power of data will bring new players onto the stage, ranging from non-FI behemoths like Google and Facebook to InsurTech startups who understand how to clean and manage large swathes of niche data. Incumbent insurers who cannot master data science will likely see themselves reduced to capital providers, competing fiercely with recently arrived third-party providers such as pension funds and sovereign wealth funds.
Claims as we know it may become a thing of the past as more and more insurance products — thanks to AI and complementing data sources — take on parametric form, making claims reporting redundant and payments are made automatically.
In conclusion, AI is likely to remove the rubber stamps from Lloyd’s of London, even removing the trading floors altogether. Thus, bringing down costs from +40% of premium revenue per transaction to no more than 0.1% or less, as is the case in modern banking.
Author; Silvi Wompa, CUO
Originally published at www.chai-uk.com.
