The processes of automation of insurance advice; analysis of advantages and disadvantages, organisational challenges, and the liability of insurance distributors.

Robo-advice and insurance distribution

The automation of advice in the financial sector, and more specifically in the insurance segment, is a reality that has been gaining traction in recent years.

Robo-advice is the end product of algorithmic activity and represents a preliminary step to the execution of insurance contracts.

In view of accelerated technological development, it is important to question whether the current insurance distribution regime already satisfactorily regulates robo-advice.   

Generally speaking, the answer is yes, although the current European insurance distribution regime regulates robo-advice mainly based on principles rather than detailed legal rules.

Indeed, the concept of insurance distribution does not distinguish whether a distributor uses conventional means or automated processes for this purpose. Similarly, the concept of advice does not distinguish the means by which advice is provided.

The principle of neutrality as to the means used to distribute insurance is mirrored in Recital (10) and Article 12 of Commission Delegated Regulation (EU) 2017/2359 of 21 September 2017 on insurance-based investment products.

Advantages and disadvantages

It is recognised that the use of algorithms in the provision of advice increases the scale of the service itself due to the greater predictability and uniformity it brings to the service.

In principle, faced with an identical baseline scenario, the outcome of advice formulated using an algorithm will be the same.

In fact, the greater the number of people involved in advising activity, the less likely it is that the advice provided will be the same, even in the case of identical baseline scenarios, taking into consideration human subjectivity.

In contrast, human error tends to be more limited in scope. Conversely, an error in the algorithm will mean that all cases where it has been applied will suffer from the same defect.

Organisational challenges

The quality of robo-advice will depend on the architecture of the algorithm, which although not usually created by the insurance distributor, but rather developed by a third party contracted for this purpose, is nevertheless part of the internal organisation of the distributor using robo-advice.

Therefore, the distributor must have the capacity and expertise to operate a distribution channel which uses robo-advice.

Liability

In view of the above, the question arises as to who is liable in the event of failure of the algorithm: the distributor or the developer?

It appears from the insurance distribution regime that liability is primarily borne by the distributor who uses robo-advice.

However, robo-advice operates within the limits of its software, which the distributor does not normally control or produce.

In any case, even the attribution of liability to the entity that designed the algorithm may be impacted by developments in machine learning, i.e., the algorithm may have the ability to learn without being explicitly programmed, thus escaping from the developer’s control.

In the resolution adopted on 16 February 2017 with recommendations on Civil Law Provisions on Robotics, the European Commission draws attention to the need to consider the potential liability of the algorithm's creator.

The Commission also proposed the possible creation of a compulsory insurance regime for robotics similar to that in place for car insurance.

In addition, it could be argued that, in the event of a software failure, recourse could be sought by the distributor against the creator of the robot.

Finally, in cases where the algorithm has undergone a machine learning process, one could even consider whether this does not constitute a force majeure event.

However, the definitive answer to the questions discussed above will depend on the legislative developments at the European level.

#Tech_Hub_VdA | Tech Hub by VdA - A Global Digital Impact