Regulatory impact analysis using LLMs, generative AI and vector databases

About the course

NEW course and format - 6 hours of training over two days

Large language models (LLMs) are a powerful tool for financial institutions provided users understand the limitations and implement them effectively. This course equips participants with tools to extend their boundaries, prototype ideas and deepen their understanding of how these models work to automate tasks and improve productivity.

Through the hands-on use of ChatGPT, participants will curate the right personae, embed guardrails into prompts and carefully construct the context of the LLM to improve quality, relevance and reliability of outputs. While customisation can enhance GenAI performance, successful usage relies on managing key risks such as hallucination and misquotation. Sessions will explore the uncertainty around safe usage and evolving regulatory expectations through analysation of model structure and the safeguard creation.

Although the future of AI governance and usage is uncertain, participants will have the confidence to successfully implement LLMs into their organizations.

Day 1 will primarily use ChatGPT; day 2 will prioritise Anthropic Claude. Participants should set up access to a free or paid account for ChatGPT and Claude ahead of the meeting. It is recommended to use a personal internet connection to avoid firewall restrictions. The class will work exclusively on publicly available data which we will provide for you.
 

Tutors