Survey Offers View into Bermuda Insurers' Artificial Intelligence Use

Nearly two-thirds of Bermuda commercial insurers do not currently use artificial intelligence or machine learning in their operations, with only 23.7 percent of property-casualty companies using the technologies, according to a report from the Bermuda Monetary Authority (BMA).

The BMA surveyed commercial insurers and insurance groups in late 2021 to gather information about the sector's use of artificial intelligence (AI) and machine learning (ML). The survey was meant to assist the BMA in assessing the current state of Bermuda insurers and their future plans regarding AI and ML.

The BMA report, titled Bermuda Insurance Sector Artificial Intelligence and Machine Learning Survey—2022 Report, notes that typical AI and ML capabilities include speech, image, and video recognition; natural language processing (NLP); conversational agent capabilities such as chatbots; predictive modeling; augmented creativity; smart automation; advanced simulation; and complex analytics and predictions.

The BMA's survey found that 68 percent of insurance groups reported using AI/ML systems, a finding the report said was expected given the scale of those organizations' international operations. The opposite was true for small commercial insurers, however, according to the BMA, while over one-half of large Bermuda commercial insurers also do not currently use AI/ML technologies.

Of the overall group surveyed, 38 percent of respondents reported using AI and ML systems. Of the 62 percent not currently using the technologies, 23 percent indicated they plan to implement them within the next 5 years, the BMA found.

"These survey results suggest that 54 percent of insurance groups and the broader commercial market plan on using AI/ML within the next 5 years, allowing the authority and the Bermuda insurance market sufficient time to develop a creditable and fit-for-purpose AI/ML framework," the BMA report says.

The BMA identified several top obstacles to commercial insurance companies adopting artificial intelligence or machine learning systems.

• AI/ML systems aren't critical to current business offerings
• No viable business case at present
• A lack of requisite skills and expertise to implement the technologies
• Limited budget

Asked about their concerns about adopting AI or ML technologies, the Bermuda insurers surveyed most frequently cited auditability, modeling challenges, system security, transparency, and the consistency of outputs.

Among the uses for AI and ML, cyber-security threat detection led the list of applications for Bermuda insurance companies currently using the technologies. Respondents also indicated using AI/ML technologies to enhance their underwriting and claims and risk management processes, among other uses, the BMA says.

The governance and control measures most frequently in place for existing AI and ML systems include senior management accountability, a dedicated data science team, regular validation, and steering bodies, the BMA reports. A small percentage of respondents reported using independent governance boards, external certifications, and alignment with a specific standard, according to the BMA.

Respondents indicated a number of areas of concern around artificial intelligence and machine learning in which they indicated they could benefit from further regulatory and supervisory guidance from the BMA. Among the top concerns were security/hacking risks, execution risks, accuracy of outcomes, regulation of systems in general, categorization bias, and transparency.

Most of the commercial insurers responding to the BMA survey using AI/ML systems said third-party or off-the-shelf services were their primary sources of AI/ML infrastructure, followed by in-house developments and customized third-party services.

Outsourced functions were governed through service level agreement compliance reviews, third-party risk evaluations, vendor confirmation of compliance, and, in some cases, third-party service audits, the BMA says.

Some 66 percent of respondents who are using AI or ML said they had no written digital strategy. But, of those with a deliberate AI or ML strategy, 73 percent reported that that strategy was set and written at the board level, the BMA says.

"An encouraging insight from a governance perspective is that the current controls for AI/ML digital strategies are set at all three lines of defense, including the operational and enterprise risk management levels and compliance function/internal audit," the BMA report says.

Respondents recommended that the BMA's regulatory and supervisory approach to artificial intelligence and machine learning included issuing guidance notes, best practices documents, and expanding the cyber-security rules and enhancements in the BMA's existing Insurance Code of Conduct.

"The BMA recognizes that AI/ML systems create opportunities, including discovering additional risks and perils leading to new insurance products, a more streamlined insurance life cycle brought about by increased interconnectedness, and new ways to underwrite traditional lines," the BMA report says. "Alternatively, AI/ML systems carry an increased risk profile, particularly in cyber security, data privacy, and legal and compliance uncertainty, not to mention ethical considerations and the risks of unintended outcomes."

The report notes that given the varying degree of AI and ML adoption among various entities in the Bermuda insurance market including insurance groups, large commercial insurance companies, small commercial insurers, captives, special purpose insurers, and intermediaries, it must develop a risk-based approach to regulating and supervising AI/ML systems.

"Each subsector is expected to have varying levels of adoption and impact on their respective risk profiles; therefore, the BMA is not looking to implement a one-size-fits-all approach," the report says.

The BMA also acknowledges that various jurisdictions will likely pursue their own approaches to addressing commercial insurers' AI/ML risks. Consequently, the BMA will look to facilitate a risk-based approach for insurance groups and large commercial insurers that operate internationally, looking to harmonize the various regulations those companies face.