Investment in AI is an obvious target for the insurance sector. Insurers have always been interested in technology that helps detect and prevent fraud and improve underwriting efficiency while speeding processes and reducing – or at least not increasing – costs.
But what is the reality in this highly regulated industry? A recent study about the use of data and AI in the Asia-Pacific (AP) region reveals some answers.
The state of play
The level of AI maturity in the insurance industry is relatively low in the AP region. Overall, 60% of insurers are in the bottom half of the maturity spectrum. However, 32% reported that their use of AI was functional – the top category of that bottom half.
Only 5% were still evaluating AI, with 23% saying they are fully mature and using AI to transform business models and processes. Just a small number of organizations were working with AI between those two extremes.
Study results suggest it could take a while before the bulk of the insurance sector is mature enough to benefit fully from AI. However, there is a clear expectation that there will be returns from investments in AI-based technology.
Investments generally focus on areas such as claims management and personalized customer experiences. The most common use is reducing claims fraud, but insurers are also interested in omnichannel product delivery, intelligent pricing and product customization, and insurance contract valuation.
Related: See how this Turkish insurer combats fraud and processes claims faster using SAS
Benefits and challenges
All these examples have the potential to increase profits while improving service delivery and addressing compliance requirements. The benefits from such technology investments include more automation with model development and deployment, and more streamlined and agile decision-making processes. Technology can also help insurers deliver more contextually relevant, personalized customer and policyholder engagements while improving governance and control.
There are considerable challenges to overcome before these benefits can be delivered. The first, and probably most important, relates to data.
Many insurers report that their data foundation is not centralized or optimized for the cloud. This makes it hard to train AI models, and hard to capture all the correct data to run the models. Indeed, almost half of insurers reported specific data problems, such as:
- Data quality issues that affect model accuracy.
- Insufficient metadata.
- Issues with data documentation.
Other challenges relate to the multitude of insurance industry regulations, and the lack of people skilled in working with AI technology.
Delivering the future of insurance
What do insurers need to do to overcome the challenges of implementing and using AI so they can achieve great returns on their investments? The first step is to improve data handling and management.
Historically, AI initiatives were hampered by poor data availability or fragmented and incomplete data sets. This was especially true when multiple sources of data were required. To overcome this challenge and ensure data integrity, insurers need to invest in robust data governance systems and processes.
Data integrity and governance underlie successful AI implementation and are essential given the highly regulated nature of the insurance industry. The sensitive nature of insurers’ data makes a focus on good data governance even more vital. Governing the data can also support core functions, such as reducing insurance claims fraud.
Once this data foundation is laid, insurers can concentrate on moving from a functional approach to AI toward a more strategic approach. Insurers should look to integrate AI across operations, ensuring its use aligns with key business objectives. Important areas for investment include scalable infrastructure and cross-functional collaboration.
Developing capabilities
The final area for insurers to consider is how to develop the skills required to implement AI. They will need to consider investing in data science and AI teams, either by upskilling employees, recruiting or bringing in consultants. This will ensure that they can build trustworthy AI models and maintain trust in those models on a long-term basis.
As AI and generative AI (GenAI) continue to advance, there will be more opportunities to modernize outdated systems, improve claims processes, and enhance operational efficiency. This can only happen if insurers have already built a solid foundation of data governance.
