Debate around whether artificial intelligence (AI) has met the key conditions of the Turing Test continues, and surging interest in AI has sparked a wave of new questions about its future impact on economies, markets and sectors, according to 2024 Global Manager Survey about Artificial Intelligence Integration in Investment.
This report presents the results of Mercer Investments’ AI integration in investment management global manager survey, conducted in 2024. The survey included 150 asset management managers from various asset classes.
Beinsure Media collected most responses from Mercer’s survey investment management, technology, and business development teams of asset management companies listed in GIMD. The insights gathered provide valuable information on AI adoption in investment management.
Use of AI in investment strategies
Challenges in agreeing to a definition of AI reinforce the complexity of determining exactly how managers are using and integrating capabilities. Yet, there is clear consensus among managers about what constitutes AI, with what might be termed the “core capabilities” being Generative AI (gen AI), large language models (LLS), natural language processing (NLP) and machine learning (ML) models.
Current use of AI across investment strategies and research stretches far beyond the traditional “quant” cohort. Nine out of 10 managers are currently using (54%) or planning to use (37%) AI within their investment strategies or asset-class research.
The integration of AI within investment strategies is not a new phenomenon; it is a future prospect.
Hedge funds, quantitative and systematic strategies have been harnessing the power of ML, NLP and trading-pattern recognition for many years. However, our findings demonstrate that current use of AI across investment strategies and research stretches far beyond the traditional “quant” cohort (15%–20% of respondents).
A minority of managers are deploying AI in more complex aspects of portfolio management.
Managers’ use of AI across investment research and alpha generation is largely focused on augmenting existing capabilities through the expansion of data sets and analysis and idea generation.
Just a small minority of managers report fully automated statistical, ML and deep learning (DL) models. Across all three areas, a significant proportion of current AI processes remain reliant on constant human intervention, reinforcing the role of AI and ML technologies as a supportive “tool” rather than a direct replacement for humans across the investment process - see about New Technology Trends: Big Data, AI & Machine learning.
Defining AI tools and the prevalence of ‘core capabilities’
The challenges of defining “what counts” as AI and the breadth of potential interpretation adds complexity to determining the scope of AI in both investment strategies and operations.
Merser findings suggest that managers clearly agree on what constitutes AI, suggesting that when they report use of AI, they are referring to what could be termed “core capabilities” — gen AI, LLMs, and NLP and ML models.
More than half of AI-integrated investment teams report that AI analysis informs rather than determines final investment decisions. A fifth report that AI proposes investment decisions, which investment teams can override.
AI is not as new as many people think it is
More than half of managers (54%) report current use of AI within investment strategies or for asset-class research, demonstrating the expansion of AI integration beyond quantitative and systematic managers to those running fundamental strategies - see How Can AI Technology Change Insurance Industry.
Although over a third of managers (36%) are not using AI in an investment or research context today, they are planning to do so in the future.
Just 9% of respondents have no plans to use AI for investment strategy and research purposes, emphasizing that AI integration and use-case development is increasingly the norm.
In aggregate, some 91% of managers responding to our survey are currently using or planning to use AI within their investment strategy or asset-class research. This provided a very broad starting point from which we drilled down into underlying uses and trends.
The role of AI in research and alpha generation
Although managers may interpret both aspects of investment processes (that is, research and alpha generation) and AI integration in different ways, clear trends emerge in our data.
Managers’ use of AI across investment research and alpha generation is largely focused on augmenting existing capabilities through the expansion of data sets and analysis, idea generation, and the identification of proxy signals where information may be more limited.
Enhanced data gathering, access and analysis is at the forefront of managers’ use of AI in pursuit of alpha generation, though a smaller minority of managers are deploying AI in relation to complex aspects of portfolio management.
In research and alpha generation, 40% of managers are using AI for big data analysis, which may translate to the incorporation of alternative data sets for predictive, descriptive and prescriptive analysis. Examples cited by managers include use of AI for searching archives, deriving security rankings and summarizing transcripts.
Nearly a third of managers (32%) use AI to support their idea generation, whether that means refining an investment universe, identifying new opportunities or justifying new trade ideas. One manager has trained an NLP model to categorize sentiment in fundamental analysts’ notes and predict future performance.
A marginally lower proportion (31%) are harnessing AI to identify data and signal proxies for missing information (31%).
A quarter of managers (25%) report using AI to support investment decision-making, broadening inputs to investment risk-management frameworks (21%), and portfolio construction and rebalancing (18%).
In relation to rebalancing, one manager reported the development of a random forest factor-timing model, which adjusts investment strategies based on value and growth factors.
Across investment strategy, use of AI is more prevalent in building “bottom up” views around individual security selection relative to assisting in the formation of “top down” macro perspectives.
Just 14% of managers view the use of AI applications as a default and key part of their investment process. More than half of managers (53%) use or intend to use AI applications as part of individual security selection informing a “bottom-up approach,” compared to 37% that use or intend to use AI to support the formation of “top down” macro views.