An artificial intelligence system that was fed Institutional Shareholder Services Inc.’s publicly available proxy voting guidelines, as well as activist and management statements on shareholder proposals, can effectively predict how shareholders will vote on resolutions, the co-authors of new research on leveraging AI as a proxy advisor found.
In their study, published Aug. 18, the University of South Carolina’s Matthew Souther and the University of Rhode Island’s Choonsik Lee found that an AI “for” recommendation is associated with a 2% to 6% increase in voting support from investors over that of an AI “against” recommendation, even after controlling for ISS’ recommendation.
That demonstrates that the AI system’s predictive power is a statistically significant predictor of voting outcomes, Souther explained.
In addition to this, the study found that the AI system produced recommendations that aligned 79% of the time with ISS recommendations, Souther and Lee reported in their paper available in SSRN.
“That suggests that ISS’ guidelines are very much relevant to what ISS is doing,” Souther said.
Heading into the project, Souther said that he wasn’t sure if AI would be capable of generating reasoned recommendations for shareholder proposals given the amount of processing power that’s required to analyze proposals. “There’s a lot of text that the AI has to crawl through to come to a recommendation,” he explained.
The AI system that Souther and Lee created for the study functioned as an ISS analyst tasked with making recommendations on shareholder proposals. A key aspect of the AI’s design involved matching each proposal to relevant ISS guidelines, Souther said.
Developing the AI system required a significant amount of hardware to run it and a lot of meticulous data work, but the methodology is replicable, he continued.
For the study, Souther and Lee used shareholder proposals filed in the U.S. between 2009–2021. The researchers’ study ended with proposals submitted in 2021 because ISS stopped selling its recommendations data to academics at the end of that year, Souther said.
More information:
Choonsik Lee et al, Beyond Bias: AI as a Proxy Advisor, SSRN (2025). DOI: 10.2139/ssrn.5362236
University of Rhode Island
Citation:
AI found to successfully predict proposal outcomes using ISS guidelines (2025, September 15)
retrieved 15 September 2025
from https://techxplore.com/news/2025-09-ai-successfully-outcomes-iss-guidelines.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
