Out-Law Analysis | 24 May 2022 | 3:20 pm | 4 min. read
In the wake of the Covid-19 pandemic, many industries have increased their reliance on artificial intelligence (AI).
The Irish insurance sector is no different. According to the European Insurance and Occupational Pensions Authority (EIOPA), its 2019 thematic review of big data analytics in motor and health insurance showed approximately 30% of insurance undertakings indicated that they were actively using AI in their business, while another 25% were at the ‘proof-of-concept’ stage.
It is reasonable to assume that if EIOPA's review was repeated in 2022, the pandemic's impact on insurance policy distribution and administration is likely to demonstrate a material increase on AI reliance.
In its Consumer Protection Outlook Report for 2022 (24 page / 1.56MB PDF), the Central Bank of Ireland (CBI) said that while technological advancement provides great opportunities to improve access and choice of financial services, regulated financial service providers must ensure that they pay sufficient care to also mitigating the risks of harm to customers that can arise from the use of technologies.
The CBI cautioned that where customers lack digital literacy or access to digital tools this can give rise to financial service inclusion and accessibility challenges. EIOPA has effectively echoed those concerns.
A balance needs to be struck between allowing the insurance industry to embrace AI, while at the same time protecting the interests of potential and existing customers of insurance distributors
As AI has been demonstrated to improve predication, accuracy and automation, insurance providers can expect to see related cost efficiencies. AI is currently used across all stages of the insurance value chain, including pricing, distribution and claims management.
While customers should also benefit from AI, which can provide a wider choice of insurance products, tailored insurance cover and more competitive pricing, they may be exposed to technology driven risk.
In EIOPA's recent report on the application of the EU Insurance Distribution Directive (IDD) (63 page / 1.4MB PDF), it noted that AI is increasingly used to provide robo-advice (an automated or semi-automated system) to customers wholly or partly and to charge different prices based on customers' behavioural characteristics.
At first glance, AI should deliver a win-win situation. However, EIOPA strikes a note of caution, noting that a balance needs to be struck between allowing the insurance industry to embrace AI, while at the same time protecting the interests of potential and existing customers of insurance distributors.
EIOPA said when a customer relies on robo-advice to choose an insurance product, there is heightened risk that they make an unsuitable decision, for example due to an error or a functional limitation in the robo-advice AI tool.
The IDD regime generally does not make a distinction between physical advice and robo-advice given to a customer. Where advice is given to a customer prior to the conclusion of an insurance contract, the IDD requires the insurance distributor to specify, on the basis of the customer's information, the demands and needs of that customer and to provide the customer with objective information about the insurance product in a comprehensible form to allow that customer make an informed decision. An insurance contract proposed to a customer must be consistent with the customer's demands and needs.
For competent authorities and insurance distributors active in the European Economic Area the current regulatory environment appears to present a challenge in terms of understanding how the IDD should be complied with by robo-advisers. Regulatory guidance at EU level would be welcomed.
Under the IDD, insurance distributors only need to inform customers about their identity and address and whether they provide advice about the insurance product being sold. Therefore, in certain instances a customer could be unaware that they are interacting with an AI machine tool rather than with a human.
EIOPA recommended in its recent report that the IDD should be updated to address both the opportunities and risks arising from the use of AI, namely robo-advice, fairness in pricing practices and regulatory supervision of AI. It said the current regime was not fit for the digital age and the growth of new distribution models.
The report also said that while increasingly personalised premiums present customer opportunity, they can also present customer threat. In behavioural premium pricing, risk adverse behaviour on the part of the customer is rewarded.
EIOPA said risks inherent in personalised premiums included a greater risk of uninsurability, which could lead to high-risk customers such as those with pre-existing medical conditions being excluded from the insurance market.
Insurance undertakings and intermediaries operating in Ireland which have or are considering building AI reliance into their business operations should consider a number of requirements.
Under the EU (Insurance Distribution) Regulations 2018, an insurance distributor carrying on insurance distribution must ensure that they act honestly, fairly and professionally in accordance with the best interests of their customers.
Under the CBI's Consumer Protection Code, there exist general principles that in all its dealings with customers an insurance distributor must act honestly, fairly and professionally in the best interests of its customers and the integrity of the market, and that an insurance distributor must act with due skill, care and diligence in the best interests of customers.
The code also includes a general principle that in all its dealings with customers, an insurance distributor must ensure, without prejudice to the pursuit of its legitimate commercial aims, that it does not, through its policies, procedures or working practices, prevent access to basic financial services.
Where an insurance distributor has identified that a person is a vulnerable consumer, they must ensure that the consumer is provided with reasonable arrangements or assistance to support them in their dealings with the distributor.
Insurance distributors must consider the risks that are inherent in using AI tools, and put controls in place directly or through cooperation with third party AI suppliers to manage those risks and to ensure that customers can benefit from the use of technology.
While the IDD regime is perhaps no longer completely fit for purpose for use in a digital age, that does not detract from the overarching obligation on insurance distributors to ensure that they act in the best interests of customers. Before introducing AI into the insurance value chain, an insurance distributor should take steps to identify and monitor what associated technology driven risks their customers may become exposed to.
Given EIOPA’s comments on the application of the IDD, it is likely that the regime will be amended in the not-too-distant future in order to keep pace with the insurance industry's digital transformation.
The use of AI is also likely to come under enhanced regulatory scrutiny in Ireland, and the growth in reliance on AI and the potential for the occurrence of associated customer risks needs to be identified and monitored at board and senior management level within insurance undertakings and intermediaries.
31 Mar 2020