Line
AI in Text Index: S&P 500 Earnings Calls
How often calls use the words language models overuse, against their level before ChatGPT (January 2021 to September 2022 = one), trailing 90 days
1 of 3 series is above their level a year earlier
Plays the chart through its history. Pauses are where the data set a record or moved most.
The AI in Text Index counts words from published research on the vocabulary language models overuse (Kobak et al.; Liang et al.), fixed before Palanor counted anything. Not an AI detector: it measures vocabulary across hundreds of companies and never says whether any company or document used AI. ChatGPT was released at the end of November 2022. The control counts twenty ordinary business words the same way. Checked against two independent methods (Fast-DetectGPT; Liang et al.): neither confirmed nor contradicted the rise, because both proved too weak on this kind of text to tell. Details are in the methodology.
The AI in Text Index tracks words such as delve, underscore, showcase and pivotal, which appear far more often in text written with language models. This chart counts them in every S&P 500 earnings call, split into the scripted prepared remarks and the unscripted analyst Q&A, and sets each against its level before ChatGPT. The control line counts ordinary business words the same way, so a rise in the first two lines is a change in word choice, not in how long or wordy calls are.
What this chart reads
- AI in Text Index: S&P 500 earnings calls, prepared remarksas “Prepared remarks”latest 2026-10-22
- AI in Text Index: S&P 500 earnings calls, analyst Q&Aas “Analyst Q&A”latest 2026-10-22
- AI in Text Index: control words, S&P 500 earnings calls, prepared remarksas “Ordinary business words (control)”latest 2026-10-22
Sources
AI in Text Index: S&P 500 earnings calls, prepared remarks. Source: Palanor calculation from earnings-call transcripts received through a licensed data provider. Calls only: one call per company per calendar quarter, S&P 500 members on the reading date (dated index membership), trailing 90 days. Prepared remarks and Q&A are split at the operator's first call for questions; the split was checked by hand on 12 calls and holds on about 94% of calls (the rest count only in the whole-call series). A value is published only when at least 150 companies have a call in the window.
AI in Text Index: S&P 500 earnings calls, analyst Q&A. Source: Palanor calculation from earnings-call transcripts received through a licensed data provider. Calls only: one call per company per calendar quarter, S&P 500 members on the reading date (dated index membership), trailing 90 days. Prepared remarks and Q&A are split at the operator's first call for questions; the split was checked by hand on 12 calls and holds on about 94% of calls (the rest count only in the whole-call series). A value is published only when at least 150 companies have a call in the window.
AI in Text Index: control words, S&P 500 earnings calls, prepared remarks. Source: Palanor calculation from earnings-call transcripts received through a licensed data provider. Calls only: one call per company per calendar quarter, S&P 500 members on the reading date (dated index membership), trailing 90 days. Prepared remarks and Q&A are split at the operator's first call for questions; the split was checked by hand on 12 calls and holds on about 94% of calls (the rest count only in the whole-call series). A value is published only when at least 150 companies have a call in the window.
Charts are drawn from Palanor's signal library and recomputed from source data on every view. Transformations (changes, indexing, year-on-year rates) are Palanor's.