However, in a world where data is everything, the potential of these new technologies is undeniable. This is especially important for the insurance industry which is particularly dependent on the quality of the information it possesses about its customers. In fact, for the first time in history, the technological advances in AI are pushing companies to rethink the 300-year-old insurance sector at all levels, from business models to value chain, and from customer interaction to process automation.
The use of notions such as “artificial intelligence” and “data-driven” is becoming more and more frequent in the discourse of CIOs or CTOs of all the main insurance giants. In fact, according to a survey by Gartner, leveraging AI across the insurance value chain is their top priority.
So we’re at a turning point in the history of the insurance industry but how far has the revolution come? Since the ethical use of AI has been put in question in various industries, what are the implications in the insurance sector? Furthermore, how can an insurance company minimise the risks of implementing AI in its value chain?
Overview of the status of AI implementation in the insurance sector
We’re witnesses to an unprecedented paradigm shift in the insurance sector where the entry of new players in the form of InsurTech companies, compete directly with massive dominating companies and accelerate their entry into the market by leveraging AI as a key competitive advantage. In addition, investment funds are not only showing interest in these new players but despite the circumstances of 2020, they have invested in them more than ever in recorded history. During the last quarter of 2020, the share of agreements in seed funding grew by 57%, returning to pre-Covid-19 levels, and half of the transactions have been in the insurance distribution sector.
However, according to NTT Data’s AI and smart data maturity framework which starts from stage 1 (incipient) to stage 5 (enhanced), the insurance sector is only at an average rate of 2.7 (getting closer to the tactical stage) whereas others, such as tech giants (4.6) and Banking (3.7) show higher levels of maturity. By falling behind, global insurance players have made it possible for InsurTech startups to capitalize on new business models and improved customer experiences. These new players commit and deliver faster claim payments, greater price transparency and on-demand policies, while decreasing the cost and the resources required.
Nevertheless, large insurance enterprises have vast resources at their disposal and they’re putting them to use. Many have invested heavily in positioning themselves as early adopters by hiring highly skilled teams, setting-up AI Labs, Centres of Excellence or hubs at a corporate level. These centres were vital in the first phase of launching the first proofs of the concept of “learning by doing” and finally, integrating or building the required data foundations.
Nowadays the market is going through a second phase, focused on scalability and monetization. The challenge is that leveraging AI at a large scale requires that the key pillars become AI-driven so that insurance companies manage artificial intelligence and data product life cycles at scale to boost monetization.
The technical and functional aspects of how AI technology is implemented is key, but given the legal implications and overall sensitivity of obtaining and manipulating personal information, a responsible and ethical use of AI is of utmost importance.
Ethical implications of using AI
While many InsurTech companies pride themselves on their closeness to the customer, their transparency, and always having the customer at the center of their business, when does “close” become “too close”? Companies today gather more data on their customers than ever before and with the help of AI, they automate many processes, but what happens when situations like discrimination and accountability issues occur because of AI?
Some uses of AI technologies such as machine learning are objective, such as documenting a damaged vehicle, but what happens when a piece of ‘manufactured information’ that the algorithm learns to associate with certain identities affects people of a certain race or ethnicity? Discrimination and bias are the most common concerns when dealing with the ethical side of AI. Managing both personal and behavioural data on a large scale, insurers need to deploy mechanisms meant to identify and mitigate data proxies and bias, and define a clear strategy to provide their stakeholders an explanation of AI models on areas such as claims management or underwriting.
Another threat to the responsible usage of AI is defining how to hold accountable self-learning algorithms. When given access to information, do algorithms have the capacity to both understand and question it, and then to properly weigh up the answers? This is feasible with hand written algorithms, but how effective can this be when self-learning algorithms are being deployed? The issue of decision making also comes to mind as we’re going through a transition from largely human decision making processes, to largely algorithmic ones. It’s difficult to assign accountability in such cases.
Key learnings in implementing AI successfully at a company level
As the race to lead the insurance market in the near future has begun, we take a look at which best practices might make the difference between success and failure:
- The transformation needs to be managed holistically, at a company level: Appointing a chief data officer and building a team of data scientists and engineers is not enough for a successful implementation. Without transformational and ambitious changes at a core level are not made, the transition to AI has very few changes of success.
- Manage and challenge the status quo of a traditional insurance company: The typical insurer has gone through different mergers and acquisitions processes and is now in the middle of an operational excellence transformation to significantly reduce its workforce in order to be more efficient and competitive.
- Create an independent team to manage the implementation: by having autonomy but also an objective overview, an isolated team can deliver better results and deliver on the goal of being able to share with the market and the shareholders that the company really invests in this market trend.
- Establish a global, coordinated, and ambitious organization chart and operating model: to guarantee the whole company is acting under a data-driven approach, leveraging all the synergies
- Embrace new use cases: powered by cutting-edge capabilities with a shorter time-to-market.
- Think out-of-the-box: insurance companies generally focus on how leveraging AI and data-driven approaches affect their current business model and value chain, but the real transformation lies in how the business model and the value chain evolve thanks to new AI capabilities.
Change is definitely here and applying AI and smart data across the insurance value chain will continue being, without a doubt, one of the main strategic priorities for the industry in the coming years. Its implementation improves customer interaction, automates processes and decision making procedures, decreases costs and allocated resources, and these are only a few of the benefits of adopting new technologies.