IVASS has launched a survey aimed at understanding the use of machine learning algorithms by Italian insurance companies and Italian branches of non-EU insurance companies, which must complete a specific questionnaire by 30 September 2022.
On June 7, 2022, the Italian Insurance Supervisory Authority (“IVASS”) issued a press release – together with a Letter to the Market – by which it informed the market of the launch of a survey aimed at obtaining information on the use of machine learning algorithms (i.e. i.e. complex algorithms capable of finding solutions by learning from input data and based on experience, without the use of mathematical models and predetermined equations) by insurance undertakings in the customer relationship. Algorithms other than Machine Learning and those which, although using this technology, are dedicated to the management of internal business processes are expressly excluded from the survey.
The survey – which follows work already carried out by EIOPA on the use of Big Data in the insurance industry related to motor and health insurance and on certain aspects of Artificial Intelligence and Machine Learning for supervision aspects – aims in particular to understand the use of these artificial intelligence algorithms in key processes, such as customer profiling, policy pricing, claims management, use of chatbots for service customer and cross-selling.
Questions contained in the survey include aspects related to governance of machine learning algorithms (including internal policies, risk management and internal staff training), customer disclosure and consent, security of algorithms, equity and outsourcing and information on the main algorithms used and their application.
The survey is aimed at Italian insurance companies (90) and Italian branches of companies from third countries (4), which are required to complete online, via the LimeSurvey platform, a specific questionnaire before September 30, 2022.
Specifically, the aforementioned entities must provide IVASS, within 10 days, with a reference contact and an ordinary email address to the dedicated mailbox, to which an email containing the link to the questionnaire form will then be sent, along with the corresponding filling instructions. .