Matthieu Caillat
Group Chief Technology & AI Officer, Chief Executive Officer of AXA Group Operations
AXA is a responsible insurer and our approach to AI is no different. We commit to developing and using AI systems in a safe, ethical and human-centric manner to drive business transformation, providing value to our customers, employees and other stakeholders.
AXA is guided by robust AI policies, governance and risk management frameworks. We use the following Responsible AI principles as the north star to lead our AI implementation and usage.
Purpose
Our AI Systems’ purpose, success metrics and implementation are focused on the creation of value for customers, employees and society, all while taking into account potential challenges, aligned with AXA’s core values: Client first, Integrity, Courage and One AXA.
Technical Robustness & Safety
We develop, select and use AI systems that meet an appropriate level of performance, security and reliability.
Non-discrimination & Fairness
We implement measures and safeguards to ensure we use AI systems in a manner that enables fair treatment and avoids unlawful discrimination and harmful bias.
Transparency & Interpretability
We inform our customers and our people when they are directly interacting with an AI system. The processes supporting decisions involving AI systems, and their limitations, are documented to allow for traceability, transparency and explainability.
Human Agency & Accountability
We implement measures and processes for appropriate human oversight and controls of our AI systems. We leverage internal AI governance and accountability frameworks to clarify responsibilities and enable effective oversight. We ensure our AI systems are designed to support and empower individuals, augmenting human capabilities and decision-making.
Privacy & Data Governance
The protection of our customers' data and our people's data is a key focus. We build data privacy and data governance into our AI projects from the outset. We use AI in accordance with applicable privacy laws and implement appropriate data governance with respect to the data we use.
Sustainability
When designing and using AI systems we consider the effects on the environment, with the aim to minimize the impact.
From principles to practices, we operationalize our Responsible AI Principles through the ongoing development of our overarching AI Governance roll-out initiative which notably covers the following aspects:
We have established governance bodies, notably the Group AI Governance Forum, to steer and ensure our Responsible AI Principles and relating policies are adopted and implemented across AXA’s entities around the world.
We provide AXA entities with guidelines, resources, expert reviews and best practices to accelerate local deployment and foster collective literacy. We harmonize AI risk management through our AI Risk Library and support Responsible AI adoption with tools such as our Fairness Compass, which helps identify fairness metrics and control bias in AI models, and our AI Carbon Footprint Calculator, which measures the CO₂ emissions of our GenAI models. We promote Machine Learning Operations (MLOps) to industrialize good AI practices, and our Group GenAI Model Advisory body regularly assesses the performance and robustness of Large Language Models for use within AXA.
We equip our employees with the training and tools to build confidence and skills in the responsible, ethical, and efficient use of AI through our Tech, Data & AI Academy. From Generative AI to ethical and sustainable AI, the learning programs offer training modules for the responsible use of data and AI.
AXA works closely with regulators and leading research centers on responsible AI and effective governance of AI in the insurance industry. AXA has established partnerships with leading academic and research institutions and works with world class universities such as Sorbonne Université and EPFL Lausanne. We have joined the Stanford University's institute for Human-Centered AI (HAI) affiliate program with research streams focusing on Large Language Model agents and the business impact of the AI revolution. In addition, alongside other stakeholders - industrial partners, consultancies, schools, and technology companies - AXA contributes to Impact AI, a French think-tank dedicated to ethics and responsible AI.