Case study: Generative AI enabled financial decision-making

Generative AI offers new ways to revolutionise data handling and decision-making in the global financial industry and new opportunities for UK’s finance related sectors.

The research question

The financial service industry is the flagship sector and the engine room to drive the UK’s growth and accounting for 12% of the UK's total output. To accelerate progress in this area we need collaborative and multidisciplinary research that brings together expertise across AI, data management, finance, mathematics, decision science, risk management and natural language processing.

A major obstacle to making this work in real-life is the challenge of gathering, preparing and updating financial datasets for training Financial Generative AI models. They rely on the availability of timely, data with many dimensions of correlations including news, company announcements, financial reports, economic analysis, market prices and video content.

The way data is consumed by such large models poses significant challenges in curating and managing data. In seeking to address these challenges, this project holds the potential to create important, practical vector data management solutions for the real-world of financial decision making.

An artistic represenation of business people interacting with digital interfaces, charts and artificial intelligence
Image credit: Getty images/elenabs

Project aims and objectives

This project aims to research and create a complete, real-time, all-in-one heterogenous data management solution for Financial Generative AI models. The project is being done in partnership with the global leader, Abrdn Investment, and the Centre for Investing Innovation.

GAIL seedcorn funding will provide an opportunity for the research team to develop modern data storage systems and financial risk models that consider both time and place, with advanced AI features.

Objectives

  1. To create a library of financial datasets for training large financial models bringing together data sources from more than 50 countries.
  2. To create a single data model that can store and query different types of data from various sources in real-time, helping to make well-rounded decisions.
  3. To develop a proof-of-concept for managing heterogeneous data for financial models and analysing the strength of potential models.
  4. To create a proof of concept for handling different types of data for financial models and to evaluate the effectiveness of possible models.
  5. To demonstrate the practical application of the proof-of-concept for assets risk management with Abrdn Investments.

Implications

Generative AI presents exciting opportunities to develop effective financial risk management models through real-time analysis and predictions of turbulent market conditions. Having enhanced data-driven financial risk assessment solutions will further improve risk management strategies by better predicting market risks with responsible investment decisions. Ultimately, this project has an ambitious goal of developing cutting edge data management infrastructure combined with the latest generative AI technology to tackle pressing global financial challenges of trustworthy and sustainable financial investment management. The project is partnered with UK financial sector leaders and aims to maintain UK’s international leading position in investment management.

Research leads

  • Tiejun Ma - Professor of Financial Computing
  • Yang Cao - Lecturer in Data Management
  • Shay Cohen - Reader in Natural Language Processing
  • Miguel de Carvalho - Professor in Statistics and Co-Director of Edinburgh Centre for Financial Innovation
  • Sotirios Sabanis - Professor of Stochastic Analysis and Director of Centre for Investment Innovation

This article was published on