Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative AI on Wall Street -- Opportunities and Risk Controls

Jackie Shen

Abstract

We give an overview on the emerging applications of GenAI in the financial industry, especially within investment banks. Inherent to these exciting opportunities is a new realm of risks that must be managed properly. By heeding both the Yin and Yang sides of GenAI, we can accelerate its organic growth while safeguarding the entire financial industry during this nascent era of AI.

Generative AI on Wall Street -- Opportunities and Risk Controls

Abstract

We give an overview on the emerging applications of GenAI in the financial industry, especially within investment banks. Inherent to these exciting opportunities is a new realm of risks that must be managed properly. By heeding both the Yin and Yang sides of GenAI, we can accelerate its organic growth while safeguarding the entire financial industry during this nascent era of AI.

Paper Structure

This paper contains 21 sections, 1 equation, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. To Write or Not To Write
  3. The Yin and Yang of GenAI
  4. Opportunities of LLM-Based GenAI
  5. Summarization and Consolidation
  6. GenAI-Assisted Meeting Minutes
  7. Thematic Learning and Consolidation
  8. Structuring the Unstructured
  9. Unstructuring the Structured
  10. Coding-Free Code Generation
  11. Putting It All Together - Agentic Networks for Automated Workflows
  12. Risks and Controls of GenAI
  13. Hallucination - Risks of Fabricated Outputs
  14. Brainwash - Risks of Malicious Data Feeds or Manipulation
  15. Harmful Gossips - Risks of Breaching Information Barriers
  16. ...and 6 more sections

Figures (3)

  • Figure 2: LLM-based GenAI can replace pre-structured GUIs or Apps to accommodate unstructured but often meaningful requests from clients or users. Returned results can also receive bespoke rendering according to the request types, e.g., using tables, bar or pie diagrams, or XYZ plotting.
  • Figure 3: An ultimate solution to facilitating unstructured services is to utilize and deploy multiple GenAI agents, each of which is specialized to a specific workflow task. This naturally gives rise to the so-called Agentic Networks. Each agent can still invoke pre-coded software or connect to APIs if needed. Furthermore, a realistic Agentic Network could also employ an iterative error-correction mechanism in order to accurately respond to clients.
  • Figure 8: LLMs like Gemini demonstrate the remarkable capability of planning and thinking. (Notice here in Step 1.2, Gemini has turned to its own resource to actually check out the price. In real and autonomous agentic environment, it will have to rely on the identified tool in Step 1.1.)