Source: Barrios, Campbell, Johnson & Liu (2025) — “Signals or Smoke? The Determinants and Informativeness of Corporate AI Disclosures”
The part that bothers me is how calm the language sounds. It makes the risk feel distant, when the real problem is already sitting inside today’s systems and decisions.
Since ChatGPT launched in late 2022, the volume of AI-related language in corporate filings and earnings calls has exploded. Companies that had never mentioned artificial intelligence before suddenly discovered it was central to their strategy. This paper asks a pointed question: does any of that actually mean something? The distinction matters because corporate disclosure is a space where strategic language is common. Companies have incentives to signal capability, competitiveness, and innovation — regardless of whether their actual AI deployment justifies those claims. The researchers call this the “signals or smoke” question: are AI disclosures genuine signals of underlying activity, or are many of them performative? What the paper finds is that it’s both, but in systematically different ways depending on the context. AI disclosures that accompany actual capital allocation — investment in AI infrastructure, changes to workforce composition, partnership announcements — tend to be informative. They predict future performance and correlate with actual operational change. Disclosures that appear in isolation from operational changes tend to be less informative and more likely to reflect bandwagon effects. The researchers find that firm characteristics matter too.
In plain English, that is why the result matters beyond the chart. It changes where people should look, what they should question, and which comfortable assumption probably needs to be retired.
For leaders, the lesson is simple: if the risk timeline changes, the attention timeline has to change too. Waiting until everyone agrees it is urgent is usually how organisations arrive late.