Deep Learning in Banking

Chapter 6Generative AI and Large Language Models

Summary

After reading this chapter, and going through the exercises, you will be able to:

• Understand the structure of modern large language models (LLMs).
• Identify the opportunities for LLMs in banking and the pitfalls in their deployment.
• Recognize the licensing issues that may arise in using LLMs, and choose the best one for your project.
• Use code to deploy open-source LLMs for banking applications.

Out of every single application of artificial intelligence (AI) in this century, large language models (LLMs) have captured the imaginations of users, regulators, and leaders worldwide the most. While credit scores and marketing models are ubiquitous and have been influencing behavior for decades, and while most people are now used to app filters and other day-to-day tools that are powered by different machine learning and AI applications, it is this new family of models that seem to be all-knowing and conversational that is revolutionizing how people relate to AI. This has not come without fears, many of them justified. From job losses to climate consequences to detrimental development in young minds, this new technology has quickly permeated into people’s lives at a speed that many of us have rarely seen before. This chapter focuses on this technology within banking, attempting to ground its real potential for our purposes, and discussing where the pitfalls arise when developing these systems in a real-life financial environment. ...