When JP Morgan Asset Management told investors last month it would stop using third-party proxy advisers for US voting and instead rely on an artificial intelligence (AI) platform, it signalled a possible turning point for the industry.
The decision comes amid heightened scrutiny of the role of proxy advisers and a belief among some managers that AI is now advanced enough to handle the scale and complexity of proxy voting.
Advocates say AI can take the more routine tasks away from human desks, allowing stewardship teams to focus on decisions that matter. Critics worry it could do the opposite by accelerating a conveyor-belt process and weakening stewardship.
Will Goodwin, co-founder of Tumelo, which launched its own AI proxy-voting research platform last month, said the investment industry faces a chronic capacity crunch and that “the workflow is broken”.
“Stewardship teams get slammed because thousands of meetings land in a tight window, and research can arrive very late, sometimes only days before a meeting, leaving little time to engage companies,” Goodwin said.
Proxy adviser backlash
In January, after JP Morgan AM’s move became public, the director of the US regulator’s investment management division presented AI as a “compelling opportunity” for managers’ proxy voting.
In a speech, Brian Daly backed AI proxy-voting research as a potential response to what he described as an ecosystem dominated by a proxy-adviser “oligopoly”, as well as the operational burden created by “vote everything” practices.
Daly said AI tools could help firms process proxy materials and generate recommendations more efficiently.
The use of AI is the latest development in a gradual reconfiguration of the proxy voting model as political and regulatory scrutiny has intensified in the US, according to Jane Sadowsky, a veteran director with three current corporate board roles.
The other changes include the larger proxy advisers beginning to move away from benchmark off-the-peg recommendations and large investors building up internal stewardship capabilities.
Last month, Wells Fargo’s wealth and investment management division ended its relationship with ISS, saying it would instead use an internal voting system.
Writing in Fortune magazine last month, Sadowsky said AI will change some things in proxy voting, but not others.
“AI promises what proxy advisors once did: scale, consistency, and speed,” she wrote.
“But AI does not eliminate judgment. It relocates it.”
The shifts this will bring, she writes, mean that boards need to rethink governance questions and engagement.
Lindsey Stewart, director of stewardship research and policy at Morningstar Sustainalytics, said that the introduction of AI in proxy voting decision-making should be seen as a positive but comes with some risks.
“The challenge for the asset managers using AI is to demonstrate that they have a good handle on the risk management over the process and that they can defend their decision-making with evidence,” Stewart said.
JP Morgan AM’s new AI-powered tool will handle the data collection, research and recommendations that its proxy advisers previously handled.
In a memo, the asset manager said the tool would “extend the high bar of independent analysis that our portfolio managers, research analysts and stewardship teams have always applied to every vote”.
Rob Hardy, stewardship veteran and former head of investment stewardship for JP Morgan AM, says AI could help with the “boring and routine stuff” that characterises a lot of proxy voting.
However, he also worries that introducing AI-assisted proxy voting research “could be an excuse to cut costs from stewardship teams”.
Cost, he wrote in a blog, was “the same exact reason asset managers flocked to proxy advisers in the first place all those years ago”.
“And we’re back to the age-old issue of having to do proxy because regulators tell you to, but nobody really wanting to pay to do it well.”
Speaking during a webinar about the best use of AI tools in data collection, analysis, and engagement, director and North America head at consultancy Canbury Insights, Gregory Elders, said that AI should “defragment” documents and connect signals across multiple sources such as proxy filings, investor materials and NGO content to create evidence-backed rationales and engagement questions, with humans firmly in the driving seat.
“Good engagement is ‘art and nuance’, not something to delegate to a model,” said Elders.
Keeping tabs
Hardy suggests asset owners should ask asset managers to clearly explain how they use proxy advisers and AI, specifying which votes are automated and which are always reviewed by humans.
They should also clarify what controls prompt a human review for significant or unusual proposals, preventing automatic decisions.
Ultimately, credibility depends on clear explanations and accountability, said Hardy.
This includes providing solid reasons for controversial votes, managing conflicts effectively, naming who is responsible when issues arise, and showing that AI has improved the quality of stewardship.
