AI expansion will bring both opportunities and threats
November 4, 2024
On 30 October 2024, we were pleased to attend the Brazil-UK Insurance Forum: Cooperation on Climate Change, Innovation and Infrastructure, hosted by the ABI and its Brazilian counterpart, CNseg.
The forum brought together industry and government experts to consider how to address the global insurance protection gap in the run up to COP30. Miqdaad Versi participated in the panel: “Innovation Trends for the Insurance Industry: AI, Cybersecurity, and Big Data,” alongside Richard Vinhosa (CEO of EZZE Seguros); Luke Foord-Kelcey (Global Head of Cyber at Howden Re); Caroline Dunn (UK Chief Underwriting Officer at Zurich) and moderator Roberto Santos (Chairman of CNseg).
It was great to see insurance sector specialists and politicians so passionate about the societal importance of insurance and its ability to facilitate positive change. Our key takeaways from the session are set out below.
Big data for risk assessment and predictive modelling
Big data is beginning to reshape risk assessment in insurance, enabling companies to predict and mitigate losses with greater accuracy. Some examples of how this is being put into practice were highlighted as part of the session.
Firstly, insurers are continuing to expand their use of multiple data sources as the capability of pricing systems to ingest a wealth of data has increased. For example, Zurich’s ‘fire score’ combines 27 external feeds with existing data to evaluate the likelihood of a building catching fire in the next 12 months and builds this directly into pricing models.
Secondly, it was noted that insurers are using data to pre-emptively reduce risks, with a positive knock-on impact on claims and premiums. Howden, for example are actively tracking a number of external data feeds and alerting their clients to potential cyber vulnerabilities and how to respond.
Finally, we heard how increased access and use of data can also smooth the claims process and provide a better service for customers. A practical example of this in Brazil saw insurers using a combination of vehicle and weather data to pay out motor claims in a flood region without ever having to see the vehicle itself.
Data protection and privacy
Insurers hold a significant amount of data on their customers making it a high-risk sector when it comes to cybercrime. As the volume and complexity of data has increased, so has the threat from hackers. Both the UK and Brazil have strong cybersecurity frameworks which are constantly being improved and updated. However, cybersecurity software can only do so much. Individuals within firms are the greatest cybersecurity threats. There is a reason why PICNIC (‘Problem In Chair, Not In Computer’) is a common acronym within the cyber risk community. Howden stated that strong internal training could reduce cyber threats by 95%. Insurers and their clients cannot stand still, however, as how we use and access data evolves over time and cybersecurity measures must evolve in step. AI is also driving change here, as hackers are increasing their use of AI to stage more complex attacks. Cybersecurity firms also need to embrace the capabilities of AI for defence.
The market and demand for cyber insurance
Panellists noted that cyber insurance is a relatively new space, with demand stretching beyond just indemnity with insureds also requesting a higher degree of pre- and post-loss services compared to other classes of business. With big data and AI expanding in many industries the need for cyber insurance is only going to increase.
Many current cyber clients are those in higher-risk industries, such as healthcare. Cyber insurance has scope to grow both through existing industries buying more cover and ‘lower-risk’ industries realising that they have a need for cover. Cyber insurance will develop into a necessity, not a luxury.
As a relatively new line of business, pricing in cyber is difficult due to lack of data, particularly on claims. As the market expands and more years of data are added, pricing will improve and more insurers will move into the space, providing the capacity to meet the demand.
Ethical implications of AI usage
The final point of discussion focussed on potential issues with AI; in particular, the implicit biases of models due to the dataset they are trained on. The panel discussed the risks of ‘hyper-personalisation’, in which algorithms may deem an individual to be ‘higher-risk’ based on personal characteristics leading to increased risk of ‘uninsurable’ segments of the market. Despite forthcoming regulatory frameworks from the EU designed to prevent harm in this space, panellists noted that insurers are at high risk of unintentional discrimination based on underwriting analysis.
The Oxbow Partners View
The rapid advancement of AI in the insurance industry has brought about exciting changes; claims and vulnerability predictions are more accurate than ever and increasing demand in the cyber insurance space points to more opportunities for growth in the market. Insurance has always been at the forefront of driving innovation and change in the wider market. New technologies present novel risks that need to be mitigated by insurance before the technology can be widely adopted.
Insurance adds value to society through transferring risk – enabling growth and innovation – but increasingly it will go further than this. Big data and AI can enable the insurance industry to proactively prevent losses from happening in the first place; the ideal outcome for customers, insurers and society as a whole.
About the authors
Hannah Huang, Lilith Mayer, Harry Schlote, Ava Spencer-Jones and Nicholas Njopa-Kaba are Consultants at Oxbow Partners.
