How Equifax Drives Responsible AI Innovation: A Q&A with AI Governance and Model Risk Management
AT EQUIFAX, WE RECOGNIZE THE IMPORTANCE OF ARTIFICAL INTELLIGENCE (AI) - AND ARE INSPIRED BY THE CAPABILITIES IT BRINGS TO STRENGTHEN OUR BUSINESS and accelerate the value of our differentiated data. Our employees around the world are constantly exploring new solutions that give customers the deeper insights they need to move people forward, faster.
We have been driving responsible AI innovation for more than a decade, with more than 1,000 global analytics professionals anticipating the evolving challenges that our customers and consumers face. As we innovate not just for today, but for the future, we are laser-focused on the importance of ongoing governance and oversight to ensure that we use AI Systems in a transparent, trustworthy, fair, explainable and secure manner.
Ritu Sharma, Senior Vice President of Global AI Governance and Model Risk Management, operates at the forefront of EFX.AI innovation. Her team is charged with exploring responsible and ethical ways to implement EFX.AI solutions. Here, she shares her passion for technology and commitment to AI governance.
How is EFX.AI used, and what governance processes are in place?
Equifax has led the way towards an industry standard for explainable AI, or xAI. Explainability remains one of the central commitments of EFX.AI. It is important in ensuring that the decisions assisted by AI are fair, ethical and understandable. Ultimately, explainability provides accountability in the use of AI, providing specific detail on what informed a decision. This is unlike today’s generative—or “black box”—AI, where humans cannot currently determine why AI generated a result. Equifax analytics experts are working on innovative solutions to solve this challenge.
Our Global AI Governance and Model Risk Management team works to align our solutions with the compliance and regulatory requirements in every geography in which we do business. We also ensure that the data science is accurate and that models are ethical, valid and deliver the expected results by maintaining a clear process and verifying that we are adhering to that process. My team makes sure everything comes together in a responsible manner and in a solution that customers find valuable.
These guardrails are important because our teams are constantly innovating. We validate new models and attribute suites (which are compilations of hundreds of calculated columns that are used as the foundation for models) and are continuously monitoring existing models to make sure they are healthy and performing as expected.
Can you explain the process and how your team is involved?
My team — along with many business stakeholders — are involved in developing models. Governance is critical and our team puts global standards in place that ensure the safety and soundness of our models. We review new use cases involving EFX.AI so that we understand the risk portfolio. This is a formal process with detailed validation and model development checklists. There are multiple, formalized key control points in which we are involved in the review.
We perform this due diligence before anything goes into production; in this way, governance is built into the process and the stakeholder teams are involved throughout. Once a use-case is approved, our data scientists outline a production plan with other stakeholders that considers the legalities, compliance and impact on customers. This begins with an understanding of the model requirements, design and timeline and continues through steps like testing, analysis, and correction all the way to implementation, auditing and documentation.
Our formal charter for AI governance defines the vision, mission, and objectives, and establishes the operational capabilities of the AI Governance Program Steering Committee, and the operation of the AI Governance Program globally. We have oversight over what Equifax teams are creating with EFX.AI technology, making sure Equifax is meeting its obligations for using AI Systems appropriately, responsibly, and in compliance with applicable laws and regulations, as well as our enterprise AI standards—and ensuring the people leading development of our solutions are aligned and educated.
How does Equifax stand out in terms of EFX.AI?
Equifax has driven responsible AI innovation for more than a decade. As of October 2024, Equifax has more than 100 approved patents supporting our approach to AI. Our Responsible AI Policies and Principles help ensure that we consistently and appropriately design, implement and use AI Systems for approved use cases. This knowledge and history of EFX.AI innovation provide a foundation from which we are able to responsibly explore and test new use cases.