This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: Companies that use AI tools to write their earnings presentations are sounding increasingly alike, a new study finds. Jujun Huang, an assistant professor of management information systems (MIS) at Binghamton University's School of Management, recently teamed up with professors from Kennesaw State University and Arizona State University to publish a paper titled "The Penalty of AI-assisted Disclosure: Language Convergence, Innovation, and Market Reaction." The work is published in the journal Academy of Management Proceedings.
The team studied a large sample of transcripts from public companies' quarterly earnings calls. Each consisted of two main parts: the prepared presentation and a live Q&A session. They found strong similarities across the presentations because of the widespread availability of AI.
The live Q&As, on the other hand, tended to be distinct due to the unplanned nature of the questions and responses. "This difference is important because it is consistent with our explanation: AI is more likely to influence communication that can be prepared, revised and polished in advance than communication that is delivered spontaneously," Huang said. Huang also found that stylistic convergence is associated with more negative initial market reactions.
When the language becomes more convergent, investors may perceive the disclosure as more boilerplate and infer that it contains less new or firm-specific information. As a result, the disclosure may appear less informative, leading investors to react less favorably. "I have always been interested in how technologies such as AI change not only how we work, but also how we communicate," Huang said.
"AI-assisted writing tools can make communication clearer and more efficient, but they may also smooth away the details and individual voice that make a message informative and credible. That tension is what draws me to this research." The next steps for Huang are to extend this research to other forms of corporate disclosure, such as annual reports, press releases and sustainability reports, to see whether similar patterns emerge. She hopes that her research can help clarify the long-term effects of AI-assisted communication.
"I do not see this research as an argument against AI," Huang said. "AI can be a very useful tool. The important question is how to use it thoughtfully." Jujun Huang et al, The Penalty of AI-assisted Disclosure: Language Convergence, Innovation, and Market Reaction, Academy of Management Proceedings (2026).
Extract — continue reading at the source.
