Evaluating AI for Finance: Is AI Credible at Assessing Investment Risk?

TL;DR

This paper evaluates the credibility of AI systems in assessing investment risk, revealing significant biases and inconsistencies across different user demographics, which challenge their regulatory compliance and reliability.

Contribution

It provides a comprehensive analysis of multiple AI models' performance in risk assessment, highlighting biases and regulatory issues in current AI systems used for finance.

Findings

Models show significant variance in risk scores based on user demographics.

GPT-4o assigns higher risk to Nigerian and Indonesian profiles.

No model maintains consistent scores across regions and demographics.

Abstract

We assess whether AI systems can credibly evaluate investment risk appetite-a task that must be thoroughly validated before automation. Our analysis was conducted on proprietary systems (GPT, Claude, Gemini) and open-weight models (LLaMA, DeepSeek, Mistral), using carefully curated user profiles that reflect real users with varying attributes such as country and gender. As a result, the models exhibit significant variance in score distributions when user attributes-such as country or gender-that should not influence risk computation are changed. For example, GPT-4o assigns higher risk scores to Nigerian and Indonesian profiles. While some models align closely with expected scores in the Low- and Mid-risk ranges, none maintain consistent scores across regions and demographics, thereby violating AI and finance regulations.