The Equation: How ChatGPT Revealed ‘Bloat’ in Corporate Disclosures
- August 25, 2023
- CBR - Artificial Intelligence
In a sample of US companies’ annual reports from recent years, the management discussion and analysis section contained almost 20,000 words on average. Not all investors have the time and expertise to make sense of so much text, which raises the question: Does the MD&A section need to be so long? Chicago Booth PhD student Alex G. Kim and Booth’s Maximilian Muhn and Valeri Nikolaev used ChatGPT to summarize MD&A disclosures, as well as transcripts of earnings conference calls, and their findings suggest that the answer is no. The researchers measured text “bloat” by calculating the difference in word count between the original document and what ChatGPT determined to be its optimal summary length—which trades off “signal” (relevant) and “noise” (redundant) words. The disclosures analyzed were bloated by about 70 percent, they find, indicating a high amount of relatively uninformative content. To learn more, read “ChatGPT Could Help Investors Make More Informed Decisions.”
Illustration by Peter Arkle
For the first time, researchers have come up with a figure—$22.5 million for the median trademark.
What’s a Trademark Worth?Stanford’s Anat Admati describes how we can build a better financial system.
Capitalisn’t: The Capitalisn’t of BankingLuigi Zingales and Tyler Cowen debate the price of tech giants’ dominance.
Should We Care If Google Is a Monopoly?Your Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.