April 1, 2024

How Investors Are Utilizing Artificial Intelligence

Generative AI tools show promise but, for now, are mostly used for clerical tasks.

Reported by Matt Toledo

Art by Klaas Verplancke

Artificial intelligence and machine learning are not new to an investment manager’s toolbox. However, the onset of generative AI tools like ChatGPT-4 and others have changed how many view and use the technology, making some investment managers rethink how to best use these tools in their operations.

Clerical and Operational Work (For Now)

Last year, several public and corporate pension fund CIOs told CIO about their plans to use generative AI in their operations. Many of these CIOs use ChatGPT and other AI tools for clerical purposes, like summarizing meeting notes, although some said they want to explore using AI to make investment decisions.

Michael Oliver Weinberg, an adjunct professor at Columbia Business School, says AI is being used in a “human plus machine” context, with AI applications still needing human supervision. A large language model could summarize a meeting or even draft a document, but “the human has to go through it and make sure it’s logical, it’s not missing anything. Those are the concerns and issues.”

One pension fund CIO said last year that by using AI tools to summarize meeting notes and upload them to the fund’s customer relationship management platform, it saves the fund’s investment staff hundreds of hours per year of manually transcribing and uploading transcripts. Another CIO said their fund explored using machine learning to analyze documents, including quarterly earnings reports.

PGIM is a firm that also uses artificial intelligence for operational purposes. According to a spokesperson, PGIM’s and innovation team created an internal generative AI sandbox using OpenAI large language models. according to the spokesperson.

According to a report from financial data technology provider Alveo, the “Leveraging AI in Financial Data Management Industry Survey Report 2024,” 41% of senior decisionmakers at financial firms say their organization has deployed AI across multiple business functions, while 97% say their firms are using AI to some degree.

“The use of AI across financial services organizations today is still in its infancy,” wrote Mark Hepsworth, Alveo’s CEO, in the report. “While a significant proportion of firms have dived in to deploying AI strategically across their entire operations, others have looked to dip their toes in the water at this stage and are waiting to see how the market evolves in the coming months and years. We would expect to see the proportion of firms deploying AI extensively over time to ramp up as the market matures.”

Asset management firm Schroders is pioneering AI in daily use for its employees. The firm in 2023 rolled out an AI assistant called ‘Genie,’ a secure, internal AI assistant to every employee. The currently has more than 1,800 weekly users across the firm, who use the tool for investment research, translations, software development and other work, according to Kevin Burson, head of digital, Americas, at Schroders.

Genie has become an invaluable tool for Schroders engineers. Burson indicated it has identified errors in code and offered fixes, even finding solutions to problems, saving the time of developers.

“Giving our employees this hands-on experience with these tools serves to upskill our staff as well,” Burson said in a statement to CIO. “As they use the tool, they are learning prompt engineering skills to ensure LLMs respond with relevant outputs. As we make the tool more robust with the deployment of more complicated use cases, we expect this to make for a more knowledgeable and skillful user base.”

Making Investment Decisions

Can an AI tool make investment decisions on behalf of a fund? It is a question being explored by many asset owners.

Earlier this year, the Massachusetts Pension Reserves Investment Management board began an exploratory program to develop an internal AI model that would be trained on decades of the fund’s data. The goal, in the long term, , is to explore how the tool could make investment decisions for the fund.

AI tools are also being used to capture and track trends across multiple industries, wrote Dennis Walsh, co-head of quantitative investment strategies at Goldman Sachs Asset Management, in the firm’s public pensions quarterly report for the third quarter of 2023.

“Many of the parts of our investment process that employ AI techniques and alternative data seek to capture the latest trends and drivers within the market impacting companies.” Walsh wrote.

Not a New Tool for Investing

Then again, some funds, especially those with a quantitative approach, have used AI tools for investment decisions for years.

Many funds have used and continue to use tools that predate today’s LLMs, Columbia’s Weinberg says. These investors use tools such as support vector machines or neural nets.

“On the liquid equity side, I would argue AI has been making investment decisions for some time,” says Weinberg, who co-founded MOV37 LLC in 2016. MOV37 was the first multi manager to focus exclusively on managers using machine learning alternative data.

“For some years, managers have been using AI to make investment decisions in liquid equities, commodities futures—that actually has been going on for a bit,” Weinberg says. “Definitely they’re usually systematic strategies run by quants or Ph.Ds., where there’s some element of machine learning and or alternative data. Now, with the advent of [large language models], they will increasingly be part of those strategies. Even if you go back to the last five to eight years, we started looking at these managers which were using [natural language processing], which is really just a precursor of an LLM. An LLM is just a large NLP model, so I would argue that’s been done for some time.”