Artificial intelligence (AI) can unlock and analyse data in ways that people alone simply cannot. Hélène Stanway and Ashish Umre of XL Catlin, discuss how AI can help risk managers to do their jobs and insurers to design risk transfer solutions to meet their needs.
How can AI be harnessed to help risk managers to do their jobs?
Hélène Stanway: The use of AI has thrown up some fascinating insights into risk trends and we believe it will really help risk managers focus on those areas that they need to target.
We have been running several proofs of concept – involving a start-up, a mid-sized company and one very large company – which has enabled us to compare technologies and strategies. One of the huge benefits of AI for our industry is that it can help to access data that currently is locked in documents or spreadsheets, for example, as well as being able to look at a span of data that, physically, a human cannot.
The proofs of concept were all exploring technologies that we believe could be of huge benefit to risk managers and enable them to better do their jobs. One of these, for example, involved examining thousands of loss reports to identify and understand trends that a human reading these documents could not identify – and thus giving us intelligence that we simply couldn’t otherwise gain.
Using that augmented information can give risk managers the information they really need to, for example, prioritise their site visits.
How are insurers using AI to find solutions to help risk managers?
Ashish Umre: Companies are building and using more flexible data platforms that enable them to more effectively use the vast amounts of data – both internal and externa – that is now available. By helping us better understand the vast landscape of risks our clients are facing, at a level of granularity that simply isn’t possible using the linear models of today, AI has the potential to unlock new risk profiles, business models and market opportunities.
Partnering with start-ups, in order to explore what is possible, means we are learning fast and by doing.
What role can AI play in insurance?
Hélène Stanway: Insurance is driven by data. As insurers, we collect data on the risks our clients face and we then analyse that data to design risk management and risk transfer solutions. But those risks are changing and AI can help us access new types of data and to combine data from different sources and analyse it much faster than humans ever could.
AI tools can help us discover patterns and, more importantly, give us meaningful insights into data. That being said, AI is just one part of our toolkit. We believe, for instance, that Blockchain has the potential to transform the industry.
In what way are you investing in AI and its development? What are you expecting from your partnerships with AI start-ups?
Ashish Umre: We are investing in emerging technologies in a number of ways. Firstly, our venture capital arm, XL Innovate, invests directly in innovative technology-driven start-ups – some of which are powered by AI. In addition, XL Catlin invested significantly in talent by setting up an internal innovation team, called Accelerate, last year.
Accelerate is now in collaboration with start-ups. A great example of this is our work with Cytora.
Cytora specialises in AI and machine learning and has created a risk engine that can capture open source data and turn it into risk intelligence. In a nutshell, the engine can be set up to trawl data from the web, news articles and even public government datasets, and process it using algorithms to understand the likelihood of future claims, risk profiles and quality of risks. This will help our underwriters in their work.
Are there limits to the development of AI in insurance? What do we need to be mindful of?
Ashish Umre: Insurance is – and always will be – a people business. While AI algorithms and machine learning are extremely powerful, and allow us to process data much faster and more accurately than a human could, we need to be very smart in how we use it.
Our primary goal is to use AI to enhance and support our insights and decision making. The provenance of data, data quality, ethics and privacy are important factors to consider when designing systems that learn from data.