Source: Jia et al. (2025) — “Corporate Responses to Generative AI: Evidence from Conference Calls”
The thing I like about this paper is that it complicates the easy story. It is not simply “AI good” or “AI bad.” It is more like: AI changes the work, and then the real problem moves somewhere less obvious.
When ChatGPT launched in November 2022, corporate responses varied widely — from immediate public enthusiasm to carefully measured silence. This paper tracks what executives actually said about generative AI in earnings conference calls after the launch, and maps the patterns of who talked about it, when, and with what tone. The surge in managerial discussions of generative AI was immediate and substantial. Most discussions were positive in tone and included specific action initiatives — not just acknowledging the technology but describing what the company was doing about it. This is somewhat different from previous technology waves, where corporate acknowledgment of new technologies often lagged their actual deployment. The cross-sectional variation in who discussed AI is informative. Larger firms showed the largest increases, consistent with having more communication infrastructure and stronger incentives to signal AI capability to investors. Growth firms and younger firms were more likely to discuss it than mature, value-oriented businesses. High-tech industries dominated — particularly computers, software, and electronic equipment. The utilities, chemicals, and energy sectors showed the smallest increases. Managers of firms with greater innovation intensity measured by patent filings, higher cybersecurity exposure, and greater labour exposure to AI were more likely to increase AI discussions.
My takeaway: AI rarely removes the hard part. It relocates it. The task gets faster, and then humans still have to decide what matters, what to trust, and what could quietly go wrong.