For an ethical & trustworthy AI in the European insurance sector

About the event
Following the publication of EIOPA’s thematic review on the use of Big Data Analytics in motor and health insurance on May 2019,  EIOPA created a Consultative Expert Group on Digital Ethics in insurance aiming at enabling a wide debate amongst stakeholders on the opportunities and challenges arising from technological innovations such Artificial Intelligence that go beyond purely regulatory or supervisory considerations. The stakeholder group has developed the report titled Artificial Intelligence (AI) governance principles: towards an ethical and trustworthy Artificial Intelligence in the European insurance sector, which was published on June 2021. 

The report developed by the stakeholder group seeks to enable stakeholders from the insurance sector to harness the benefits arising from AI, while at the same time addressing the challenges in a proportionate manner. The proposed framework recognises the freedom of insurance firms to select the combination of governance measures that better adapts to their respective business models, and more particularly to the concrete AI use cases that they aim to implement, but at the same time highlights those areas requiring special consideration so as to promote trust in the use of AI by insurance firms. 

Objectives
Stakeholders from the Consultative Expert Group on Digital Ethics in insurance will present the content and context of the AI Governance Principles report, and gather feedback from stakeholders.
The event offers participants the opportunity to exchange views with EIOPA and other stakeholders from across Europe.

Duration 
3 hours (distributed in three different days)

Registration to this event is currently closed.

Agenda

 Panel discussion: AI fairness and ethics

The advent of new technologies such as Artificial Intelligence (AI) are enabling opportunities for future growth and development in the insurance sector. The benefits arising from AI in terms of prediction accuracy, automation, tailoring products and services or cost reduction are remarkable. However, there are also concerns about the impact that the increasing adoption of AI could have on the financial inclusion of groups of protected classes or vulnerable consumers or on our society as a whole. 
•    What does fairness and ethics mean in an insurance and AI context? 
•    How do insurers use AI across the insurance value chain and how does it impact consumers? 
•    What is a biased dataset and what measures can be put in place to prevent AI systems from producing biased outcomes?

Panellists: 
•    Lutz Wilhelmy, Actuarial Association of Europe
•    Olivier Jérusalmy, Financial Inclusion Europe
•    Martin Mullins, University of Limerick

Moderator: Tim Shakesby, EIOPA 

22/09 11:00 - 12:00 Panel discussion: AI transparency and explainability

The provision of transparent information to consumers in order to allow them to make informed decisions is one of the cornerstones of insurance legislation. Other stakeholders such as auditors, the board of insurance undertakings, or supervisory authorities also need to be provided with relevant information. The issue of transparency and explainability in the context of Big Data and AI is becoming increasingly important; in addition to the increasing use of new sources and types of data, certain AI systems are labelled as “black-boxes” because the rationale of the outcome/prediction of the system can be difficult to explain.
•    How do AI systems differ from traditional models used in insurance (e.g. generalised linear models) in terms of transparency and explainability?
•    What measures can be adopted to increase the transparency and explainability of AI systems and the accountability of its users?
•    What type of information / explanations need to be provided to the different stakeholders in insurance?
Panellists: 
•    Esko Kivisaari, Actuarial Association of Europe
•    Jasper De Meyer, BEUC
•    Chris Holland, Loughborough University
Moderator: Tim Shakesby, EIOPA 

24/09 11:00 - 12:00 Panel discussion: AI Governance (focus on human oversight, data management and record keeping, and robustness and performance)

The Solvency II Directive requires insurance and reinsurance undertakings to have in place an effective system of governance which provides for sound and prudent management of the business. In an AI context, this can be achieved in different ways, such establishing adequate levels of human oversight throughout the AI system’s lifecycle. Sound data management and record keeping can also help reinforce the accountability of firms. Insurance firms should use robust AI systems, both when developed in-house or outsourced to third parties, taking into account their intended use and the potential to cause harm.
•    What roles do the different staff members of an insurance undertaking play in the oversight of an AI system?
•    Why is record keeping of AI systems important for ensuring adequate accountability and redress mechanisms?
•    When do AI systems need to be retrained, recalibrated and revalidated?
Panellists: 
•    Pedro Ecija Serrano, Grant Thornton
•    Marcin Detyniecki, AXA
•    Mirko Kraft, Coburg University of Applied Sciences and Arts
Moderator: Tim Shakesby, EIOPA