Bringing together expertise from the investment industry and multi-disciplinary researchers, the project aims to create a generative AI-powered research companion, that will combine an enhanced large language model with statistical inference to support abrdn’s investment teams. The focus will be on expanding both the breadth of securities that abrdn’s investment teams can review, as well as the depth of quantitative analysis that they are able to draw from, tailored to the abrdn investment style.

Unlike some of the other AI technologies being deployed by asset managers in investment decision-making which focus on identifying specific risks or analysing prior decisions, the tool will support abrdn’s research teams by applying their process in a scalable way across a broader range of securities to deliver more timely and richer insights. Crucially the tool will go beyond delivering an enhanced version of a customised large language model by adding a ‘statistical brain’.

Researchers from the Schools of Informatics and Maths at the University of Edinburgh are exploring how existing large language models (LLMs) might be adapted and enhanced to synthesise the huge amount of data needed to make investment decisions – combining public information as well as abrdn’s proprietary data. The new tool would also provide explainability of how decisions are reached.

It is envisaged as an add-on to existing team capabilities, with the reports and insights generated requiring human intelligence to be overlaid. The integration of the tool is expected to bring greater capacity to abrdn’s investments business as it will enable quicker and more informed investment decisions and will help to broaden abrdn investment teams’ coverage within and across sectors. The project will also create opportunities for students and early career researchers within industry, as well as knowledge-sharing around forecasting and prediction for academics.

Echo Yang, investment director at abrdn and the project co-lead, said: “AI has the power to make a substantial contribution to our existing investment processes but like any new technology it has to be harnessed correctly to deliver real benefits.

“By working with the University of Edinburgh’s world-class team we can bring the best academic insight, and match that with abrdn’s real-world experience of markets and investing.

“There is an ever-increasing range of data points to analyse as we constantly seek to deepen our analysis of the markets we invest in. Through exploration of using generative AI to support our investment teams, we have the potential to deliver even more value for our clients.”

The project being announced forms part of a wider relationship already in place between abrdn and the University of Edinburgh, supported by Edinburgh Innovations, the University’s commercialisation service. In April 2022, the two parties announced the creation of the Centre for Investing Innovation, a £7.5 million strategic partnership to address challenges facing the investment and asset management sector across three main areas: sustainability, thematic investing and innovating investing. The centre sits within the Edinburgh Futures Institute, a new futures-focused space for learning, research, and innovation at the University of Edinburgh.

Professor Sotirios Sabanis, director for the Centre for Investing Innovation alongside co-director Professor John Vines, said: “Developing a high impact financial decision-making model requires a collaborative effort and interdisciplinary expertise integrating, in our case, mathematical modelling, natural language processing, and insights from the psychology of human behaviour.

“As well as benefiting investors and customers, research co-created with industry like this also offers secondment opportunities for students, enhances our education programmes, creates knowledge-sharing opportunities for researchers and ensures our research is helping to meet society’s challenges.”

The joint project aims to deliver an alpha test model for trialling in the business in 2025, and a full prototype by the following year. Throughout this journey, and with any potential integration into the live research process once ready, the model will be under the supervision of abrdn’s AI Governance Framework (AIGF), which has been established to enable the company to take strategic advantage of artificial intelligence whilst making sure the risks are understood and managed.

The University of Edinburgh has recently established two AI hubs for causal AI and electronics, as well as launching the Generative AI Laboratory (GAIL) last year. The University was recently ranked first in the world for Industry, Innovation and Infrastructure (Times Higher Education Impact Rankings 2024).