Artificial intelligence (AI) identifies patterns or produces outputs from data. Machine learning (ML) fits those patterns from examples. Generative AI produces text, code or other content; fluent output is not evidence that its factual claims are correct. The assigned International Monetary Fund (IMF) October 2024 reading studies how these capabilities change capital markets. The Financial Stability Board (FSB) November 2024 reading examines their wider stability implications.
In trading, predictive models can estimate returns, liquidity and execution costs. An execution system splits an order and selects venues to reduce market impact. In portfolio management, models combine prices, filings and other information to inform allocation and risk forecasts. In compliance, pattern recognition flags suspicious transactions for investigation. Client services include document processing and customer support. Predicting a return, choosing an action and generating an explanation require different validation.
Generative models can extract information from disclosures, draft research, write analytical code and support dialogue with clients or supervisors. More autonomous systems could combine research, portfolio decisions and execution.
Common mistakes
- Equating speed with stability. Faster price discovery can coexist with simultaneous selling and reduced stressed liquidity.
- Counting vendors instead of dependencies. Different applications can run on one upstream provider.
- Waiting for harm before supervision. Risk-based oversight includes controls before deployment and monitoring during use.
Bottom line
- AI can improve research, execution, risk analysis and supervision.
- Better ordinary liquidity can coexist with fragile crowded exits.
- Supply and demand jointly drive adoption.
- Common providers, correlated models and weak governance connect firm risks.
Exam shortcut
When an AI scenario changes, trace what is shared: information, trading decisions, infrastructure or all three. Choose a control that interrupts that failure. For private-credit comparisons, write both funding shares before comparing rates. Distinguish a cross-country association from the separate cost-of-capital calculation and reject a precise growth forecast unsupported by the inputs.
The full lesson (about 3,993 words, 27 min read) adds 2 worked examples, all 6 common mistakes, a self-check, free in the app.
Learning objectives
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
Browse all free FRM Part II lessons or jump into free FRM Part II practice questions.