Financial firms must put explainability at the centre of artificial intelligence (AI) systems to safeguard trust, compliance, and risk management, warns a new report from the CFA Institute, a global association of investment professionals.
The report looks at how AI systems are becoming more complex in areas like credit scoring, investment management, insurance underwriting, and fraud detection. It stresses the importance of “explainable AI” — methods that ensure decision-making is “transparent, auditable and aligned with human understanding.”
The study lays out a framework showing how different groups such as regulators, risk managers, investment professionals, developers and even clients, each need different kinds of explanations from AI systems. It also explains the tools available to provide this clarity: some are built in from the start (“ante-hoc” methods), using simple rules people can easily follow, while others come in after the fact (“post-hoc” tools), showing what data influenced a decision or how the outcome might have changed under different circumstances, such as a borrower having a higher income.
EIOPA issues AI oversight guidance for the insurance sector
Recommendations include framing global standards for AI explanations, tailoring interfaces to different users, promoting real-time explainability and investing in human–AI collaboration. The report also points to alternatives such as evaluative AI, which presents evidence for and against decisions and neurosymbolic AI, which combines reasoning with deep learning.
“This is not about slowing down innovation; it’s about implementing it responsibly,” added Rhodri Preece, senior head of research at CFA Institute. “We must ensure that AI systems not only perform well but also earn the trust of those who rely on them.”
“AI systems are no longer working quietly in the background; they are influencing high-stakes financial decisions that affect consumers, markets, and institutions,” commented Dr. Cheryll-Ann Wilson, CFA, the report’s author and a senior affiliate researcher at CFA Institute. “If we can’t explain how these systems work – or worse, if we misunderstand them – we risk creating a crisis of confidence in the very technologies meant to improve financial decision-making.”
