A growing chasm is emerging between the fervent optimism expressed by corporate executives regarding Artificial Intelligence’s transformative potential and the tangible productivity gains currently observed in the broader economy. New research from the Federal Reserve Bank of St. Louis indicates that while AI is increasingly a focal point in corporate discussions, its measurable impact on productivity is largely perceived as a future phenomenon, even as aggregate economic data shows minimal actual improvements. This disparity underscores a modern-day "productivity paradox," where significant technological investment and enthusiasm precede demonstrable economic shifts, mirroring historical patterns seen with other groundbreaking innovations.
The comprehensive study, spearheaded by St. Louis Fed researchers Serdar Ozkan, Aakash Kalyani, and Nicholas Sullivan, meticulously analyzed a vast dataset comprising approximately 490,000 earnings call transcripts from 5,198 publicly traded U.S. companies. Spanning a quarter-century from 2000 through 2025, the research leveraged an open-source large language model (LLM) to identify sentences related to productivity, ascertain whether AI was mentioned in conjunction, and classify the temporal nature of the described gains – past, present, or future. This innovative methodological approach allowed for an unprecedented, granular understanding of corporate sentiment surrounding AI’s economic contributions.
The Rise of AI in Corporate Discourse
Before the public release of OpenAI’s ChatGPT in late 2022, mentions of AI within productivity-related discussions during earnings calls were virtually non-existent, registering near zero. This reflected a period where AI, while advancing rapidly in specialized fields, had not yet permeated mainstream corporate strategy or public consciousness as a broad-based productivity driver. The landscape dramatically shifted following ChatGPT’s debut. The accessibility and perceived capabilities of generative AI catalyzed a significant surge in corporate interest and discussion.
The research meticulously tracks this trajectory: a sharp increase in AI-related productivity sentences was observed throughout 2023, indicating a rapid assimilation of AI into corporate narratives. This initial surge plateaued in 2024, suggesting a period of consolidation, experimentation, and perhaps a cautious evaluation of initial deployments. However, the momentum regained significant pace in 2025, with AI mentions accelerating once more, ultimately accounting for approximately 15% of all productivity-related sentences by the close of the year. This pattern illustrates a dynamic corporate engagement with AI, moving from initial excitement to strategic integration and renewed, perhaps more refined, anticipation.
A Future-Oriented Outlook: Executive Sentiment vs. Reality
A critical finding from the St. Louis Fed’s analysis is the overwhelming future-oriented nature of executive statements concerning AI’s productivity impact. Across all sentences where AI was linked to productivity, an astonishing 95% referred to future gains. This contrasts sharply with non-AI related productivity discussions, where roughly three-quarters of sentences also spoke of future gains, but still leaving a significant portion for present or past achievements. The tone mirrored this pattern, with 95% of AI-related sentences describing productivity as increasing, compared to 75% for non-AI sentences. This indicates a high degree of optimism and proactive positioning around AI’s eventual benefits.
The researchers summarize this sentiment, stating that firms are "investing, experimenting and reorganizing around AI today, while the measurable productivity effects remain mostly ahead." This statement encapsulates the current paradox: companies are committing substantial resources – financial, human, and strategic – to AI initiatives, yet executives themselves acknowledge that the substantial, measurable returns on these investments are still on the horizon. This forward-looking posture is a testament to the perceived transformative power of AI, but it also highlights the lag inherent in integrating such a fundamental technology across complex organizational structures and broader economic systems.
The Broader Economic Context: Minimal Aggregate Gains
Despite the pronounced executive optimism, the broader economic data presents a more subdued picture. Data from the Federal Reserve Bank of San Francisco, cited by the St. Louis Fed researchers, reveals that utilization-adjusted total factor productivity (TFP) grew by a mere 0.07% over the four quarters ending in the first quarter of 2026. Total Factor Productivity, a crucial measure of technological progress and efficiency, reflects the portion of output not explained by the amount of inputs used in production. Its minimal growth suggests that the widespread corporate enthusiasm for AI has not yet translated into significant, economy-wide efficiency improvements.
This low TFP growth rate echoes historical instances of the "productivity paradox," first coined in the 1980s by economist Stephen Roach concerning the impact of information technology. Roach, and later Robert Solow with his famous quote, "You can see the computer age everywhere but in the productivity statistics," observed that initial waves of IT investment did not immediately yield the expected surge in productivity. Instead, it took years, even decades, for businesses to reconfigure processes, retrain workforces, and fully leverage the new capabilities before significant productivity gains became evident. The current situation with AI appears to be following a similar trajectory, suggesting that the initial phase of adoption is more about investment and adaptation than immediate, widespread efficiency dividends.
Investment and Innovation: The Seed of Future Growth
While aggregate productivity remains modest, another related study from the San Francisco Fed by Kalyani and Huiyu Li offers insights into where the investment is concentrated. Their research found that firms exhibiting positive AI sentiment during earnings calls have, indeed, demonstrated higher capital spending and research and development (R&D) growth compared to other public companies. This crucial detail indicates that the optimism is not merely rhetorical; it is backed by concrete financial commitments.
However, this combination of high positive sentiment and increased spending is not evenly distributed across the corporate landscape. It is predominantly concentrated among large technology firms that are actively building the foundational AI infrastructure – the chips, cloud computing services, and core AI models that power the broader ecosystem. These companies are the "pickaxe sellers" in the new AI gold rush, making significant investments in the underlying technology that will eventually enable productivity gains across other sectors. Their concentrated spending and R&D efforts are essential for the long-term development and deployment of AI, but the trickle-down effect to generalized economic productivity takes time.
Implications for Corporate Strategy and Workforce Planning
The Fed’s findings carry significant implications for various stakeholders, particularly for corporate strategists and human resources leaders. For businesses, the research serves as a powerful reminder that AI adoption is a journey, not an instant solution. While the long-term potential of AI to drive efficiency, innovation, and growth is widely acknowledged, the current phase demands strategic patience, continuous experimentation, and a clear understanding that initial deployments may not immediately yield the dramatic returns often touted in public discourse. Companies must invest not only in the technology itself but also in the necessary organizational restructuring, workforce training, and process redesign required to truly harness AI’s power.
For HR leaders, the research offers a critical "counterweight" to internal narratives that might prematurely treat AI productivity gains as already realized. The allure of immediate efficiency improvements through AI can be tempting, leading to swift decisions regarding workforce plans, hiring freezes, or even layoffs. However, the Fed’s data strongly suggests that such actions, built on the assumption of current and delivered efficiency gains from AI, are running ahead of what the evidence shows. Executives themselves are communicating these gains in the future tense to investors, implying that the groundwork is being laid, but the fruits are yet to be harvested.
Therefore, HR strategies must prioritize reskilling and upskilling initiatives to prepare the workforce for an AI-integrated future, rather than solely focusing on potential headcount reductions based on unproven efficiencies. Investment in training programs that enable employees to collaborate with AI tools, manage AI systems, and leverage AI-driven insights will be paramount. A cautious, data-driven approach to workforce transformation, grounded in realistic assessments of AI’s current capabilities and future potential, will be far more sustainable and effective than reactive measures based on premature assumptions.
Measurement Challenges and the J-Curve Effect
Part of the explanation for the productivity paradox also lies in the inherent challenges of measuring productivity, especially for novel technologies like AI. Traditional productivity metrics often struggle to capture improvements in quality, customization, speed of innovation, or the value of new services created by AI that may not immediately translate into higher output per hour worked in conventional terms. The full economic value of AI may manifest in ways that are difficult to quantify with existing statistical tools, at least in the short term.
Furthermore, economic historians and economists often refer to the "J-curve effect" when discussing the impact of major technological paradigm shifts. This theory posits that significant technological transitions initially lead to a dip or at least a lag in productivity growth. This is because businesses must incur substantial costs for R&D, new capital equipment, training, and the often-painful process of reorganizing work. During this period, output may not increase commensurately with these investments, leading to a temporary decline or stagnation in measured productivity. Only after a critical mass of adoption and adaptation is achieved do the sustained productivity gains materialize, causing the "J-curve" to ascend sharply. The current AI landscape appears to be squarely within the initial, flatter or descending part of this curve.
The Path Forward: Continued Monitoring and Strategic Adaptation
As AI continues its rapid evolution, the St. Louis Fed researchers have committed to continuously tracking whether the language used in earnings calls shifts from mere expectations to descriptions of realized results in the coming quarters. This ongoing monitoring will be crucial for policymakers, investors, and businesses to gauge the true economic impact of AI and to inform future strategies.
The journey of AI integration is a complex interplay of technological advancement, corporate strategy, human adaptation, and economic measurement. While executive optimism provides a powerful engine for investment and innovation, it is essential to ground expectations in empirical data. The current moment is one of significant investment and experimentation, a period of laying the groundwork for what many believe will be a profound transformation. The real test, and the true measure of AI’s impact, will come when these investments translate into widespread, tangible productivity gains that reshape the economic landscape, much as electricity and information technology did in preceding eras. For now, the future of AI productivity remains largely a promise, eagerly anticipated by those at the helm of corporate America.
