Whilst there is yet a commonly accepted definition of artificial intelligence (AI), public discourse surrounding AI is often coloured by dystopian scenarios, espoused even by famous figures such as Stephen Hawking and Elon Musk. From a technical perspective, AI is a machine-based tool that can be used to solve complex problems. This said, the current ability of AI to solve problems is restricted to specific tasks, based on training with a limited set of data, models, and algorithms (weak AI). We are still a long way from the kind of AI that can solve ‘unlearned’ problems,’ the kind capable of resulting in a drastic paradigm shift (strong AI). At the same time, the true value of AI precisely lies in the fact that its returns potential is correlative to the complexity of the problem at hand. As opposed to ‘traditional’ automation, AI will make it possible to find solutions to issues and tasks that are not understood or even known, or for which the input data is simply too large for human understanding.
For the insurance sector, AI offers incumbents the chance to automate increasingly complex processes and save costs, all the while incorporating considerably larger data volumes into decision-making processes to improve precision and quality. Yet implementing AI-based systems presents manifold challenges for insurance companies, creating an opening that fast-moving, technology-driven companies are all too ready to fill. Against this backdrop, this article will firstly discuss the opportunities AI presents along the insurance value chain, especially regarding its impact on customer touchpoints. Secondly, we will look at whether these opportunities are sufficient to turn technology-centric businesses into ‘real’ threats for incumbents.
Opportunities along the insurance value chain
An increasing number of AI-driven use cases are emerging along the insurance value chain—from product development and underwriting, sales and service, through to claims settlement— promising increasing process quality, speed, and reduced staff workload. Here, AI can be beneficial for insurers when searching for correlations in decision-making processes, where the potential decision levers are not fully known. These decisions are often based on creating scores for various problem levels, which could be solved through AI simultaneously. Take workflow management as an example: in order to preselect from incoming cases all those that could be run automatically, the nature of cases has to be defined along multiple levels, e.g. a fraud score calculated, claims cost estimated AI allows insurers to solve these problem levels simultaneously, identifying obscure and increasingly better correlations to complete the task at hand.
Opportunities at the customer interface
Beside clear potentials for increasing process efficiency and effectiveness, AI makes it possible to automate the interaction between insurance companies and customers. This interaction is highly complex due to the nature of human communication and the variety of customer needs. Here, AI can generate significant quality improvements for “simple” use cases such as automated clarifications of insurance terminology. AI agents can address questions, correct spelling mistakes, and other special cases flexibly based on natural language processing. They can also detect the customer’s mood by analysing the input text or voice entry and adapting responses accordingly. All in all, AI has the potential to offer a less rigid, simpler, and faster digital experience along the entire customer journey. Best-practice use cases include:
In a nutshell, AI shows considerable potential for incumbents. However, this is thwarted by fundamental technical and organisational requirements, shaking insurers to their very foundations: For one, IT systems need to become more flexible to integrate new solutions, data aggregated and made available, and new skills procured or taught. These challenges are creating a race against time and present new competitors with new entry points into the market.
Implications for the business models of insurance companies
According to the Big Data trifft auf künstliche Intelligenz (‘Big Data Meets Artificial Intelligence’) study that the German Federal Financial Supervisory Authority (BaFin) published in June 2018, data and AI-driven innovations are driving the disintermediation of the insurance value chain. The report proposes that traditional providers may increasingly be forced into the role of an infrastructure provider for third-party offers. It would be easier for non-insurance companies to enter the market if they possessed relevant external data along with the skills to analyse them for risk assessment. The report further argues that data and AI-driven innovations would be a major competitive lever for the battle at the customer interface.
At the heart of the insurance business model lies the task to calculate precisely the ‘expected value’ of a risk and to underwrite that risk at an attractive premium, ultimately achieving low combined ratios. This relies on a feedback cycle between risk calculation and the cost of claims:
While external data plays an increasingly important role, the risk estimations are validated by the actual incurred costs of claims and are constantly refined based on historical claims data. As a consequence, traditional insurers benefit from an extensive knowledge advantage that translates directly into profitability. What’s more, the actual costs of claims depend on many insurer-specific competencies which secure incumbents a substantial competitive advantage. One such factor is the success of cooperation with repair networks and numerous other ‘suppliers’ depending on the line of insurance, which bilaterally rely on the business volume and market power of an insurance company. Secondly, the costs of claims also depend on numerous highly specific insurance skills such as managing claims transfers between insurance companies, litigation proceedings, as well as knowledge of physical, bodily, and financial damages. These core competencies in insurance business can only be achieved by technology companies, if at all, through extensive investments in money and time. When basing risk calculations on external data alone, coupled with a lack of expertise in claims management, the cost rate for claims will be a persistent challenge for new market participants. When it comes to the core insurance business, then, incumbents seem to still stand at a clear advantage.
However, if one looks at competition at the customer interface, insurance companies appear to be in a much more perturbed position. With technology-driven best practices in mind, the insurance customer experience is still characterised by too many delays, unwanted channel switches, and incomprehensible jargon. Here, it is much easier for new providers to unseat traditional insurance companies with technological skills, as well as externally and customer data directly provided. At the same time, insurance business models that concentrate solely on the customer interface suffer from the strong pressure on margins, as demonstrated by the consolidation of brokers in price-driven markets such as the United Kingdom, or the closing down of the once hyped-up fintech company Knipp. This said, big tech companies such as Amazon are much less affected by this pressure on margins, given that they already offer a broad range of services at the digital customer interface, through which insurance products can be sold easily. With secure business models, established technological expertise, coupled with a better understanding of customer needs, big techs can slowly test the waters and expand into the insurance business through small pilot projects such as “Amazon Protect.” If incumbents fail to offer customers an up-to-date digital experience, they will struggle to compete with the big tech companies in the long run.
There is no doubt that AI brings forth a range of opportunities for insurers and will increasingly become an important competitive factor between existing insurers. Currently, an advantage in insurance and especially claims knowledge protects incumbents from new non-insurance providers. However, this will not shield them from the continuous onslaught from more digitally-driven companies in the long run. Insurers can no longer ignore the benefits of AI-driven solutions for the analysis of claims data, leveraging external data sources, as well as the technology-driven improvement of customer interactions. Speed is key, as the competitive advantages of insurers are constantly eroding.