Generative AI in Finance and Banking

Module 1: Foundations of Generative AI

Learning Objectives

• Explain in plain terms what generative AI is and how it differs from predictive AI
• Recognise the core technologies and tools driving generative AI
• Identify where generative AI already touches financial services
• Appreciate the power and limitations of large language models (LLMs)

Content

• What is generative AI? – the ‘satellite-level’ view
  • Generative AI in finance – e.g. financial report drafting, synthetic data, personalised communications, scenario stress testing, code generation, conversational financial assistants
  • Discriminative AI in finance – e.g. sentiment analysis, credit scoring, fraud detection, algorithmic trading, quantitative risk modelling, anomaly detection, portfolio optimisation, data analytics, time series forecasting
• Generative AI vs. discriminative/predictive AI:
  • Using examples from credit scoring and client briefing generation
• Generative AI vs. RPA (Robotic Process Automation)
• AI in finance training explores how LLMs are trained on selected data patterns
• Core tools in use today: such as ChatGPT, DALL·E, Claude, Gemini, along with domain-specific fintech AI platforms
• Why now? The drivers of adoption in banking and finance (data availability, cloud power, regulatory pressure, competitive advantage)
• Live Demo: Summarising a market news article
• Interactive Activity: Drafting a client-facing investment note in plain English

Module 2: Generative AI Use Cases in Banking & Finance

Learning Objectives

• Identify high-value applications of generative AI across banking functions
• Describe how AI can assist in client-facing, operational, and research contexts
• Explain how RAG improves reliability by grounding outputs in trusted sources
• Evaluate which use cases are most relevant to their own role or team

Content

• Client-facing applications: personalised portfolio updates, 24/7 chatbots
• Operations & risk management: regulatory report drafting, fraud detection explanations
• The importance of data input cleanliness
• Time savings, reduced manual effort, and the cost-benefit of generative AI
• Grounded AI outputs with RAG: combining generative AI with internal knowledge bases (e.g., compliance manuals, research archives) to reduce ‘hallucinations’ and increase trust
• Research & analysis: summarising company filings, scenario testing narratives
• Innovation & product design: product brainstorming, client persona simulation
• Interactive Activity #1: Summarising a fictional earnings call transcript
• Interactive Activity #2: Extracting compliance requirements from a mock regulatory text

Module 3: Hands-On Generative AI for Finance Tasks

Learning Objectives

• Apply prompt engineering techniques to generate effective outputs
• Adapt the style and tone of AI-generated content for different audiences
• Detect and manage hallucinations and inaccuracies
• Use generative AI to create practical outputs such as client notes and compliance drafts

Content

• Prompt engineering basics: open vs. specific prompts, role-playing prompts
• Adjusting tone: compliance-legal vs. client-friendly
• Stress testing – Red-teaming AI: testing limitations and weaknesses
• The problem with hallucinations:
  • Demo: What can go wrong?
  • Avoiding hallucinations: verifying AI outputs against data
• Demo: ChatGPT alone vs. ChatGPT-with-RAG on a regulatory document query
• Interactive Activities:
  • #1: Drafting a short client note on “The Impact of Interest Rate Changes”
  • #2: Creating a compliance reminder email for internal use
• Code Generation for Finance – Using LLMs to create small scripts for financial tasks (e.g., generating a Python snippet to pull stock data from the cloud or writing an Excel VBA macro to automate risk reporting)
• Interactive Activity #3: Use ChatGPT to write a useful Excel VBA macro

Module 4: Risks, Ethics, and the Future of Generative AI in Finance

• Identify key risks of generative AI adoption in financial services
• Discuss ethical issues such as bias, transparency, and accountability
• Summarise current regulatory perspectives and requirements
• Explain how RAG pipelines and human oversight mitigate risk
• Envision future applications of generative AI in finance and fintech

Content

• Key risks: data leakage, bias, over-reliance, hallucinations
• Ethical issues: transparency, client trust, accountability
• Regulatory perspectives:
  • EU AI Act – “high-risk” classification for financial applications
  • FCA, SEC, MAS – Emerging supervisory positions
• Mitigation and risk management:
  • Human-in-the-loop oversight
  • Measuring output quality (accuracy, bias, speed, cost)
  • AI sandboxes for experimentation
  • Vendor due diligence and audit controls
• RAG pipelines for traceability and audit-friendly outputs
• Debate: “Would you trust AI to draft a client suitability report?”
• Future Trends:
  • Generative AI in algorithmic trading, surveillance, hyper-personalised advice, and risk modelling
  • Fintech/startup partnerships with banks
• Critical success factors for generative AI in banking and finance
• Final activity: Each participant will work with ChatGPT to explore a future vision scenario in their area of work (e.g., wealth management, risk, compliance, markets) to enhance their uptake of generative AI

Requirements

This AI in finance course assumes participants have access to ChatGPT during the course and will be able to download various input files required for interactive activities via a dedicated GitHub project. For the exercise on generating Excel VBA code, it would be useful (though not essential) if participants can access Microsoft Excel during the course, with the ‘Developer’ option enabled.