Skip to content

FRM Exam Part II · The Financial Stability Implications of Artificial Intelligence

AI Adoption and Use Cases in Finance for FRM Part II

Updated 11 October 2026 · Fact-checked

Financial institutions use AI and machine learning to find patterns in large data sets. Main uses are trading, credit underwriting, risk management, fraud detection, customer service and compliance. Adoption is driven by data growth, computing power, cost savings and competition. To answer questions, match the use case to its benefit and its main risk.

Understand AI Adoption and Use Cases in Finance

Artificial intelligence (AI) is software that performs tasks usually needing human judgment. Machine learning (ML) is a part of AI. It learns patterns from data instead of following fixed rules written by a person. Generative AI creates new content such as text, code or images, often using large language models.

Banks and investment firms use these tools across the business. In trading, AI supports signal generation, execution, market making and portfolio construction. In credit, ML scores borrowers using more data than a traditional scorecard, which can widen access but can also be hard to explain. In risk management, models help with forecasting, stress testing and early warning. In fraud and financial crime, models flag unusual transactions and reduce false alerts. In customer service, chatbots and language models handle queries. In compliance, tools monitor communications and automate reporting.

Why do firms adopt AI? The main drivers are more data, cheaper computing, better algorithms, pressure to cut costs, competition from rivals and fintechs, and the chance to improve accuracy. Many firms buy AI from a small number of outside providers, which links adoption to third-party and concentration risk.

For the exam, think in pairs: use case and benefit, then use case and risk. Common risks are poor data quality, model risk, lack of explainability, bias, cyber risk, third-party dependence and herding if many firms use similar models. Reports such as the FSB's on AI and financial stability stress that current uses are often back-office and support tasks, while uses in core decisions bring higher stakes. Treat adoption as uneven across firms and functions.

How to solve AI Adoption and Use Cases in Finance questions

Use this method for any scenario question on AI use cases and adoption.

  1. 1Identify the business function in the question: trading, credit, risk, fraud, customer service or compliance.
  2. 2Name the AI or ML technique being used, such as supervised learning, anomaly detection, natural language processing or generative AI.
  3. 3State the benefit the firm gains: speed, accuracy, lower cost, wider data use or better detection.
  4. 4Find the key risk linked to that function: explainability and bias in credit, herding in trading, false positives in fraud, hallucination in generative AI.
  5. 5Check for adoption drivers or barriers mentioned, such as data availability, cost, skills, regulation or vendor reliance.
  6. 6Look for third-party or concentration clues, such as one provider or one shared model.
  7. 7Choose the option that matches both the use case and the correct risk. Reject options that overstate certainty.

Quickest way: Function, Benefit, Risk triple

When to use it: Use when time is short and the options mix several functions and risks.

  1. Underline the function in the stem.
  2. Write the likely benefit and the likely risk in two words each.
  3. Eliminate options that attach a risk to the wrong function or claim AI removes human oversight.
  4. Pick the option that is balanced and specific.

Common mistakes in AI Adoption and Use Cases in Finance

  • Treating all AI as generative AI

    News coverage focuses on chatbots and large language models.

    Fix: Remember that most financial AI use is traditional ML for scoring, forecasting and detection. Generative AI is a newer, smaller part.

  • Saying AI removes the need for human oversight

    Automation sounds like full replacement.

    Fix: Regulators expect governance, validation and human accountability. Pick answers that keep humans responsible.

  • Linking explainability concerns only to trading

    Students think of black-box trading first.

    Fix: Explainability matters most in credit decisions, where firms must justify outcomes and avoid unfair bias.

  • Ignoring third-party concentration as an adoption feature

    Students focus on the model, not who supplies it.

    Fix: Note that many firms rely on a few cloud and model providers. This ties adoption to systemic risk.

  • Assuming more data always means a better model

    Big data is promoted as an advantage.

    Fix: Data quality, relevance and bias matter. Poor or unrepresentative data produces poor models.

Worked examples

Example 1

A bank replaces its traditional scorecard with a machine learning model that uses transaction and cash-flow data for small business loans. Which is the most appropriate risk for the risk manager to stress?
A. The model may be hard to explain and may embed bias
B. The model removes all credit risk
C. The model cannot use more than ten variables
D. The model makes validation unnecessary

Show the solution
  1. Function: credit underwriting.
  2. Technique: supervised ML using wider data.
  3. Benefit: possibly better risk ranking and wider access to credit.
  4. Key risk: complex models are harder to explain and can reflect bias in the data.
  5. Option B is false because credit risk remains. Option C is false because ML handles many variables. Option D is false because validation is more important, not less.

Answer: A

Example 2

A global bank deploys the same external AI vendor model for fraud detection, customer chatbots and compliance monitoring, as do many peers. Which concern does this most directly raise?
A. Third-party dependence and concentration risk
B. Lower operational risk because outsourcing transfers all liability
C. Reduced need for data governance
D. Higher model diversity across the system

Show the solution
  1. Clue: one external vendor, used across several functions and by many peers.
  2. This points to reliance on a few providers.
  3. A common vendor failure, outage or flaw could hit many firms at once.
  4. Option B is wrong because outsourcing does not transfer accountability. Option C is wrong because governance is still needed. Option D is wrong because shared vendors reduce diversity.

Answer: A

Exam tips

  • Expect applied scenarios where you match a function to its main risk, not definitions only.
  • Look for words like vendor, shared model or same provider. They signal concentration risk.
  • Be careful with absolute words such as always, eliminates or removes. They are usually wrong.
  • Know the six uses: trading, credit underwriting, risk management, fraud detection, customer service and compliance.

Practice questions from The Financial Stability Implications of Artificial Intelligence

AI Adoption and Use Cases in Finance: frequently asked questions

What are the main AI use cases in financial services?

The main ones are trading and market making, credit underwriting, risk management, fraud detection, customer service and compliance. Each has a typical benefit and a typical risk. Learn them as pairs.

How are banks using AI in risk management?

Banks use ML for forecasting, early warning, stress testing support, anomaly detection and monitoring. These uses improve speed and pattern detection. They still need validation and human judgment.

Why is machine learning in credit underwriting a concern?

ML models can improve accuracy and use more data, but they can be hard to explain. They may also reflect bias in training data. Supervisors expect strong governance and validation.

What drives AI adoption in finance?

Key drivers are more data, cheaper computing, better algorithms, cost pressure and competition. Reliance on outside providers also shapes adoption and creates concentration risk.