Artificial Intelligence (AI) is gradually establishing itself in the technological landscape of insurance, an industry historically based on risk analysis and management. Insurance is historically an industry that requires big data analysis and processing and the advent of AI could revolutionise current organisations in a few years. To achieve this, companies are starting to develop pilot projects on different operational topics with the aim of significantly improving processes or the customer experience. Captives are currently not always the highest priority for the parent company, which favours their core business.
Today, the term AI has become a generic term to group together all new technologies without the latter necessarily being linked to AI strictly speaking. The robotisation or automation of certain processes very rarely involves the implementation of AI. Beyond the topics of process automation and cost optimisation, which do not necessarily require the use of AI, it is important to assess the challenges that captives may encounter in AI project implementation in the near future.
Integration complexity
Integrating AI into existing infrastructures is not always simple. The existing tools do not always enable the smooth deployment of an AI solution. Projects need to be planned and the introduction of new tools often marks the beginning of projects with the associated high costs.
In theory, the automation of operations (management of slips, payment of claims, production of Solvency II statements, projection of pricing, etc.) and risk management could be carried out by AI. However, to date, the expected efficiency gain still seems insufficient in our view to justify the investment in the development of efficient AI.
Training and reorientation
As soon as AI is deployed, a new challenge arises: establishing training programs that would enable employees to use this tool optimally. Employees must be trained to use the technologies, as well as to process and analyse data to enable real added value and long-term efficiency gains.
Data quality
For AI to be effective, accurate and complete data is needed. However, the quality of data may vary, posing a major challenge for AI models. Today, most captives benefit from a data history which could allow them to implement AI models. Meanwhile, the current organisation, structuring and storage of this data has not been designed for being processed by AI. One of the challenges lies in data management which involves third parties such as captive managers, brokers and even cedants. These players must operate at their own project level to transform themselves and have the capacity to provide the data expected by captives for the optimisation of their economic models.
Upcoming innovations
Technology continues to evolve which will ultimately make it possible, assuming we have suitable data to use, for example, for machine learning and predictive analysis to improve risk projection and pricing models managed by captives. Some reinsurers have already carried out studies to project climate risks and thus optimise their pricing and exposure.
Data security and confidentiality
Captives must navigate a complex and constantly evolving regulatory framework. Understanding and complying with local and international regulations regarding the use of AI is essential to avoid sanctions. Furthermore, some groups are currently asking questions about the security and confidentiality of data used by AIs on the market. Indeed, economic competition between great powers can influence the way in which data is collected and used. For example, American companies are subject to the Cloud Act, a legislation that allows American authorities to access data stored by American companies, even if that data is located outside of the United States. This raises concerns for reinsurance companies operating internationally, which must ensure that their sensitive data cannot be collected by competitors or by organisations wishing to favour their domestic players. Managing data confidentiality and protecting against economic espionage can limit the use of AI in this case.
Despite the complexity brought about by current technological disruption, the challenges associated with implementing AI are not insurmountable. With strategic planning and wise investments, captives will gradually overcome these challenges. The potential benefits of AI, such as improved operational efficiency, increased accuracy in risk analysis and better customer experience, are powerful drivers for simplifying organisations and will enable significant innovations.
By working together, captives and their partners can create robust and secure solutions, maximising the use of AI's potential while minimising risks. Long-term commitment to innovation and collaboration can turn these challenges into opportunities, allowing captives to fully take advantage of AI and benefit in the years to come.
PwC Luxembourg, in collaboration with Microsoft, this year has launched the GenAI Business Center, located at the PwC Experience Center in Luxembourg. This groundbreaking and strategic collaboration marks a significant investment by both companies in helping people and organisations thrive in the ongoing GenAI revolution. This adds yet another building block in PwC’s global strategic collaboration with Microsoft, creating scalable offerings using Microsoft Azure OpenAI Service and Copilot for Microsoft 365 to help support our clients in reimagining their organisations.
Contact us
