AI Real AI Implementation Empty AI Disclosures Investor Assessment Information Value

Signals or Smoke? The Determinants and Informativeness of Corporate AI Disclosures

Distinguishing genuine AI initiatives from marketing narratives

Authors: John Manuel Barrios, John Campbell, Ryan Grant Johnson, Christine Liu

Status: Revision Requested: Review of Accounting Studies

Research Area: Corporate Governance & Disclosure

View on SSRN Back to Research

Abstract

We examine corporate artificial intelligence (AI) disclosures to determine whether they reflect genuine operational AI implementation or represent empty marketing narratives. Using a comprehensive dataset of AI-related corporate disclosures from SEC filings and earnings calls, we analyze the determinants of AI disclosure intensity and test whether these disclosures are informationally valuable to investors.

Our findings reveal two distinct patterns. Some firms make substantive AI disclosures aligned with actual implementation—these disclosures are associated with increased R&D spending, technology hiring, and measurable operational investments in AI capabilities. Other firms make generic, boilerplate AI disclosures with minimal substantive content or implementation—these "smoke" disclosures are primarily marketing-focused. We find that markets do not fully distinguish between signals and smoke, at least in the short term, suggesting potential mispricing of AI initiatives.

Key Findings

Substantive vs. Marketing AI

AI disclosures fall into distinct categories: genuine implementation (backed by R&D spending and talent acquisition) versus marketing-focused narratives with minimal substantive content.

Disclosure Determinants

Firm characteristics such as industry, size, and prior profitability predict AI disclosure intensity, with some disclosures driven by investor relations strategy rather than operational reality.

Information Value

Substantive AI disclosures are associated with measurable operational outcomes and future stock returns, while generic disclosures show weak information content.

Market Mispricing Risk

Investors may not fully distinguish high-quality from low-quality AI disclosures in the short term, creating potential for mispricing of companies making hollow AI claims.

Research Contribution

This paper addresses an important and timely question about information quality in AI-related corporate disclosures. As AI becomes a central element of corporate narratives and investor expectations, understanding the difference between substantive AI initiatives and marketing narratives is crucial for market efficiency.

Our findings have implications for investors seeking to distinguish genuine AI leadership from hype, for regulators considering enhanced disclosure requirements around AI, and for researchers studying how firms respond to investor pressure for narrative alignment with technological trends.

Citation

Barrios, John Manuel, John Campbell, Ryan Grant Johnson, and Christine Liu. "Signals or Smoke? The Determinants and Informativeness of Corporate Artificial Intelligence (AI) Disclosures." Review of Accounting Studies (Revision Requested).
View on SSRN Back to Research