As insurance enters an execution phase of using AI and predictive analytics, will we see incremental change or exponential developments? What are the main impediments to long-term adoption?
Our CEO Richard Hartley recently offered his views on these topics during the panel discussion ‘Offering Clients What They Want Next: Predictive analytics and the impact on efficiency and delivery’, at KBW‘s inaugural Innovation in Finance conference held in London last week.
The panel was moderated by Jean Pierre Lambert, Managing Director at KBW. Other panellists included Alex Pengilly, COO at Adarga, Mireille Dyrberg, COO at Duco, and Richard Wood, Business Development Director at Synectics Solutions.
What is the current state of AI and predictive analytics in insurance? Will we see incremental change or exponential developments?
“We are seeing rapid change in SME insurance. The critical dimensions to a successful AI implementation are the availability of data and the complexity of the entity it is trying to make a prediction about,” said Richard Hartley.
“In SME insurance, risks are very homogenous and there are large pools of data that can be used to understand a risk. This makes it easier to understand the true level of exposure, which makes SME a great candidate for automated underwriting.”
For more complex prediction and assessment tasks, panellists agreed that change will be incremental. AI can provide support to humans through verifying and reconciling data but more complex tasks will remain human-driven.
What are the most exciting areas of innovation in AI and predictive analytics for insurance and finance?
The panel discussed predictive analytics as an alternative to the historic insurance business model of learning from long-term claims data. This will inevitably result in insurance becoming more competitive as AI makes it possible to establish new markets intelligently and generate growth without needing years of claims experience.
“Insurance today is very illiquid, it can take up to five years to build up claims experience you need to underwrite in a market,” said Richard Hartley. “Most insurers will only ever see at most 5% of the market, which makes it hard to diversify and grow. At Cytora, we enable insurers to underwrite in SME segments without claims experience. This enables insurance companies to scale without adding cost to their businesses.”
Panelists mentioned active learning, a type of semi-supervised machine learning, and knowledge graphs, a tool which captures a relationship between entities, as interesting areas of innovation to watch. They also discussed improving operational efficiency and empowering users as exciting areas of development.
“We want to change the business model for how banks manage data and interface with the outside world,” said Mireille Dyrberg. “The new way of structuring yourself means you can rip out operational costs, replacing legacy systems with more efficient and effective technology that does not require you to sink so much resource and money into building and maintaining it. You can make your systems so much easier to work with and really empower users.”
What are the obstacles to partnering with incumbents? And what are some of the barriers to long-term adoption of AI?
For insurtechs, gaining access to high-quality data from incumbents and the slow pace of implementation were highlighted as obstacles to successful partnerships. Established players often struggle to maintain the agile pace, constant feedback, and rapid iteration cycles that early-stage startups expect.
Legacy systems were identified as a huge barrier to the long-term adoption of AI and predictive analytics. Incumbents who have to maintain existing legacy systems find it difficult to innovate in parallel.
“Organisations are economically motivated. Insurtechs often buy into their own hype too much but don’t talk enough about economic benefits,” said Richard. “Incumbents won’t go away, and ultimately, they can innovate faster than startups can get to their distribution because insurers are great at maintaining their customer base.”