Gen AI: Turbocharging AI in Capital Markets
Exploring Use Cases, Risks and Implementation of Generative AI in Capital Markets
Abstract
With its ability to extract insights from large data sets, AI has been used in capital markets for decades, for example, in algorithmic trading. With the acceleration in the development of electronic trading, automation, and the exponential increase in data, use cases for AI have multiplied across the industry. Even more complex and less explainable models, such as deep learning models, are now frequently used for pattern recognition in AML and market analysis. Traditional models have had very targeted and narrow applications and were focused on analyzing available data. This has changed with the emergence of large language models (LLMs) and Generative AI (Gen AI). The large data sets used to train these models and the ability for the models to recognize more context from data give the models a wider application, while the generative abilities open a host of new possibilities for AI integration in Capital Markets. Many of the use cases are improvements on prior applications of AI, such as summarization, headline- and sentiment analysis, and Compliance supervision. Far from being out of the box, each model and often each use case will require fine-tuning or prompt engineering. Additionally, while Gen AI is good at many things, other, more established models (AI and other traditional statistical models) are better for other use cases. Therefore, Gen AI models will be one element within a larger AI ecosystem rather than a stand-alone system. However, creating a deeper understanding of the context of text or conversations and generating output (code, text, etc.) drives the possibility for new productivity gains in co-pilots and complete workflow redesigns. This report explores the current and future use cases for Gen AI in capital markets from summarization to a Gen AI-powered end-to-end system.
As with every opportunity, there are risks for Gen AI these include the proliferation of Bias, Hallucinations, and Regulatory concerns. Companies need to engage with this new technology to avoid being left behind and vulnerable to cyber-attacks. Cost is also a big factor as the initial data, development, and continuous processing bills can easily extend into the millions.
Despite some of these shortcomings, Gen AI has created interest across the industry, and anyone not contemplating engaging with this new technology is going to miss out on potential competitive advantage or put themselves at risk. To help participants get started with their Gen AI journey, we have set out some steps that capital markets participants can take to get started. This includes setting up business and governance frameworks, allowing your workforce to ideate use cases, and investing in infrastructure and talent.
Related Research
IT Strategy & Priorities: Capital Markets 2023 Buy side | Sell side
July/August 2023
IT Strategy & Priorities Webinar: Capital Markets 2023
July 2023
Keeping up with JPMorgan on AI
April 2023
Future of Trading Technology: The Cloud Cometh
December 2022
Technology Trends Previsory: Capital Markets Sell Side, 2023 Edition
November 2022
Enterprise Data Management visions and trajectories for Investment managers 2025 – Part 2
March 2022
Jan 2022
Optimizing Data Architecture to unleash the Power of AI
Nov 2021