Evasive top executives can now add AI to the ranks of those scrutinizing their quarterly earnings calls.
A study by researchers at the University of Münster and Washington University in St. Louis have found that analysts can use large language models to draw conclusions about when executives are not forthcoming during conference calls on quarterly results, reported the London-based investment strategist Joachim Klement.
During the study, researchers fed conference call transcripts into OpenAI’s GPT-4 Turbo to assess whether a manager’s comments were “usual” or “unusual.” If the AI detects an unusual conference call transcript, it will explain its reasoning behind it, Klement explains.
In one example, the AI analyzed the transcript and flagged that management may have been avoiding specifics despite analysts’ urging, Klement said. The AI flagged “unusualness” most often when speakers gave detailed answers and long discussions about nonfinancial topics, some of the most common ways executives deflect negative news.
Since the launch of ChatGPT in 2022, the financial industry has increasingly integrated AI into its daily operations. Earlier this year JPMorgan Chase revealed an AI tool that helps interpret Chairman Jerome Powell’s ambiguous statements. However, it has not yet been proven that AI makes a significant difference in stock picking.
Ultimately, researchers believe AI could help analysts predict the market by gaining a better view of earnings calls.
“The stock market reacts negatively to unusual financial communications with increased trading activity,” the researchers wrote in the study’s abstract. “This reaction is further amplified when multiple dimensions of unusual communication are identified for a company.”
Klement notes that with the launch of OpenAI’s GPT-4o in May, analysts will also be able to upload the audio of the quarterly earnings call to the LLM or provide it during the live stream.
While analysts often notice unusual comments themselves, technology allows them to simultaneously analyze multiple conference calls on quarterly earnings from different industries.
“Personally, I’m excited about it, but as a senior executive I probably wouldn’t do it because chatGPT makes it much harder to get away with deception and distraction,” Klement wrote.