Views on AI have evolved over Pensions & Investments’ coverage of institutional use of technology. Back in 2021, adoption of AI among institutional investors was low, said Thomas Kim, then-CEO of the financial technology software provider Enfusion.
Now, the California State Teachers’ Retirement System, West Sacramento, plans on exploring AI in its 2025 fiscal year. While it’s still doing research, the $337.9 billion public pension fund will consider what role AI will play in its innovation and sustainability goals, according to a presentation for its July 24 offsite meeting.
Other investment professionals have also been wrestling with how generative AI is being used responsibly, not only in their investment practices but in their holdings.
During the 2024 proxy season, AI-related proposals were filed by the trade union center AFL-CIO and activist investor Arjuna Capital at names like Apple and Meta Platforms.
But shareholders have also been examining AI use beyond the technology sector, especially at companies using AI to innovate in their practices, said Beth Williamson, vice president, director of sustainable equities research and associate portfolio manager at Calamos Investments, which had $38.2 billion in assets under management as of June 30.
For instance, Deere & Co. has been investing in and integrating AI into its farm equipment to help farmers increase their yield, reduce insecticide use and become more efficient.
“That’s an application that is huge, especially if we’re going to go into the sustainable investing space,” she noted. “Here’s a name that you never really think of in the AI discussion, but as investors, we’re looking beyond the Magnificent Seven (and) the IT names to find these opportunities.”
On that end, AI-related proposals were filed by the $270.5 billion New York City Retirement Systems as well as AFL-CIO at media companies including as Netflix and Warner Bros. Discovery.
These proposals sought the establishment of board committees to oversee AI and reports on the use of the technology, or more specifically, how corporations spread misinformation and disinformation created by generative AI.
Parnassus filed a proxy proposal at Verisk Analytics that asked the insurance data analytics provider for governance information in regard to decision making, particularly around discriminatory impacts on rural populations.
“It’s really important that the algorithms aren’t making decisions that worsen, create disparity or introduce discrimination when you’re looking at data — sometimes nontraditional data — to make sure insurance decision making around claims processing, underwriting or pricing,” Macindoe said.
Parnassus withdrew the proposal after the company agreed to release that information by the end of the year, she noted.
Overall, proposals on AI received low shareholder support, with one submitted by AFL-CIO asking Amazon.com to form an additional board committee to oversee the technology receiving 83.3% votes against it, including from CalSTRS. But the proposal at Netflix, which asked for AI ethical guidelines, did receive a strong 43.2% support, making it one of the top 10 investor-supported proposals during the season.
Still, with this ESG proxy trend expected to continue, Calamos’ Williamson said that “it’s important that we go back to board and executive oversight for their leadership because you can have a policy … but if the company is not living that truth, then it’s just word salad.”
“That’s where investors are going,” she added. “What are companies doing to evaluate, or what are companies doing to improve their corporate governance structure with a focus on AI safety and ethics … How many of the independent directors are there that have this AI expertise?”
Investors also worry about deepfakes, which are “deliberately used to deceive, distract (or) defame,” Macindoe said.
States such as California and New York have introduced laws on the use of AI, particularly in regard to their influence on political elections. Laws may imply penalties, but whether jail time or financial fees will discourage unlawful behavior is in question, said Macindoe.
Another concern is around “hallucinations,” or when AI gives inaccurate answers or misinformation.
“We as investors cannot afford to have two plus two equal five ever,” NCIF’s Narain said. “Portfolio analytics and asset allocation have to be accurate.”
NCIF’s AI tool doesn’t provide investing advice, but it can inform decisions, Narain noted.
Through machine learning, access to data and analytics can be democratized for nontechnical “people to scrutinize,” he noted. “I don’t think the human mind is going away, not for a long time. The human mind has to be able to get that information and make an analysis out of that.”
Institutional use cases for AI and ESG
Industry professionals told P&I that AI can parse through large volumes of financial data, including a company’s regulatory filings, corporate calls and financial news. The technology also has room to innovate in the ESG space, said Gabe Rissman.
As an undergraduate student studying data science at Yale University, he led and engaged in shareholder activism through the Dwight Hall Socially Responsible Investment Fund. The student-run fund currently manages $200,000 in assets on behalf of the university’s investment office.
Today, Rissman is the co-founder and president of YourStake, a wealth management data platform with an emphasis on values-based investing. His team integrated generative AI into their service after realizing in 2023 the technology’s capabilities corresponded with increasing client demand.
For sustainable data specifically, AI can sift through sustainability reports to pull information on water consumption or management diversity, he said. YourStake’s AI tool allows users to upload a portfolio and share what values they care about, such as water protection or human trafficking. The AI tool can tell the investor where its portfolio is matching values or lacking.
Among use cases at asset management firms, Rothko Investment Strategies offers bespoke ESG investment solutions for its emerging and international equities funds.
The independent unit of the $49 billion Mondrian Investment Partners uses large language models to extract data from “vast amounts of text information to drive a deeper understanding about controversies in the ESG space,” said Daniel Philps, head of Rothko.
Additionally, Man Numeric has been using ChatGPT-based technology in climate modeling for physical risks in its portfolios.
AI has helped with reading company reports and thinking of actions it can take to mitigate climate change-related damages in its portfolio with less staff involved, said Matt Goldklang, climate scientist at the quantitative equity subsidiary of $178.2 billion Man Group. The technology is “really just becoming a very helpful assistant,” he told an April 17 panel at the BloombergNEF Summit in New York.
NCIF’s Narain said the nonprofit’s AI tool was created out of the need for data and measurement in the context of the Greenhouse Gas Reduction Fund, a $27 billion program launched by the U.S. Environmental Protection Agency to mobilize capital to projects meant to curtail the effects of climate change.
Responding to questions in plain English, the platform has the ability to run trend analyses of demographics in a specific county or the pollution in a neighborhood.
From there, an investment professional can do asset allocation and scenario analysis. Using the example of a $50 million portfolio that invests in low-income neighborhoods focused on whole energy-efficient technologies, NCIF’s tool can solve for the impact of the technologies, the number of jobs that could be created, the amount of carbon dioxide emitted — or the market implications if more capital was allocated to the portfolio.
The final part of the process — which the nonprofit is still developing — measures the overall impact of the decision, or verifies it. Narain has showcased the tool to community development financial institutions, financial advisers, asset owners and managers — and even the EPA.
AI is going to be here for the long term, and embracing the technology will move the needle further, he added.
“Might as well,” Narain said. “I would love to convene a group of people who will talk about it and say which pain points can be useful. More than anything else, you might think about this as a call to action.”
