Financial services have used analytics and machine learning for decades. Generative AI is adding new uses.
Common Applications
- Fraud and financial crime detection: transaction monitoring and anomaly detection.
- Credit risk: scoring and underwriting models.
- Customer service: assistants answering account and product questions.
- Compliance: reviewing communications, summarising regulations, preparing reports.
- Operations: document processing for loans, claims and onboarding.
- Investment research: summarising filings and news.
Specific Risks
- Fairness: credit decisions must not discriminate.
- Explainability: customers and regulators may be entitled to reasons for decisions.
- Model risk management: validation, monitoring and governance expectations from regulators.
- Accuracy: errors in advice or information can cause financial harm.
- Security and fraud: criminals also use AI, including deepfakes.
Good Practice
Keep humans accountable for significant decisions, document models thoroughly, test for bias, and involve compliance early.