AI Transforms Finance, Unleashing New Waves of Risk

At a recent international financial forum, a senior official from the People's Bank of China addressed the profound integration of artificial intelligence with the financial sector. Lu Lei, a Deputy Governor, highlighted that the current wave of AI, driven by large language models and intelligent agents, is reshaping financial services at an unprecedented scale.

The Flip Side of Innovation: Opaque Algorithms and Model Hallucinations

Lu pointed out that alongside tremendous efficiency gains, this technological shift introduces novel and significant challenges. Two core risks stand out: the "algorithm black box" and "model hallucinations."

  • The Algorithm Black Box: While AI systems can learn and parse complex patterns of human behavior, their internal decision-making logic and reasoning processes remain largely inscrutable to humans. This opacity complicates oversight and makes tracing the root cause of errors or biases exceptionally difficult.
  • Model Hallucinations: AI models can generate outputs that appear coherent but are factually incorrect or logically flawed—a phenomenon known as "hallucination." In finance, this could lead to faulty credit assessments, erroneous market forecasts, or misguided trading decisions, potentially resulting in substantial losses.

The "Asymmetric Understanding": A Fundamental Threat to Stability

"The core issue lies in an 'asymmetric understanding' between humans and machines," Lu elaborated. Humans struggle to fully decipher AI's decision-making chain, while AI continuously learns from human behavior. If this cognitive gap is not properly managed, it could embed latent systemic risks within financial markets, potentially rendering some traditional risk-control measures less effective.

Forging a Collaborative Safety Net for the AI-Finance Era

Confronting these challenges requires open collaboration, not isolation. Lu expressed a clear commitment to strengthening regional and international cooperation. Key collaborative focus areas include:

  • Jointly researching the characteristics and transmission mechanisms of new financial risks born from AI technologies.
  • Exchanging practical experience in managing risks across the entire AI model lifecycle—from development and deployment to operation and decommissioning.
  • Sharing effective practices for enhancing algorithm transparency, mitigating biases, and improving model robustness.

The ultimate objective is to construct a resilient, adaptive safety framework that evolves in lockstep with the financial sector's digital and intelligent transformation. This endeavor transcends pure technology; it necessitates an upgrade in governance systems and regulatory capabilities to establish a trustworthy and secure foundation for the integrated future of "AI + Finance."