Strategic AI Underuse: Effort Incentives, Displacement Risk, and Organizational Responses
Stefano Staffolani · Metroeconomica · 2 Nov 2026
Where this sits in the debate
- Will AI actually boost productivity and growth? · its finding: it depends (why studies differ)
- Will AI destroy more jobs than it creates?
Direction of the finding as read from the abstract by an AI model (AI-written).
Abstract
ABSTRACT This paper studies how workers' employment concerns can affect the effective use of Artificial Intelligence (AI) tools after adoption. When productive use of these tools raises workers' perceived risk of displacement, employees may optimally reduce productive effort, trading current earnings against employment security. The baseline represents the first implementation period, when neither side has yet observed behaviour and production under the new technology. The firm evaluates adoption using the observed pre‐AI effort level, while workers forecast the production scale from the most recently observed aggregate output. Strategic underuse can therefore partly or fully offset the productivity gains expected from AI tools and make an initially attractive technology unprofitable after implementation. After implementation, the firm observes the effective effort response, while its organizational options depend on whether it can diagnose the incentive mechanism behind that response and on the cost or feasibility of workforce adjustment. The firm may choose technological reversion, labour hoarding, workforce restructuring through voluntary separations, or employment‐preserving contracts based on negotiated working‐time reductions. Robustness benchmarks show that the displacement‐induced incentive for underuse can persist when firms and workers form more forward‐looking expectations. A Monte Carlo calibration illustrates that all four organizational responses can arise as profit‐maximizing outcomes.
About this publication
Unranked journal. A journal that is not on the economics or computer-science rankings this guide uses. That says nothing about one paper, but no ranking vouches for the journal’s review.
- Published
- 2026 (recently)
- Citations
- 0 (OpenAlex)
0 citations recorded by OpenAlex — too few 2026 papers in this collection for a fair comparison.
Reaches a different conclusion
Papers on the same question whose finding points the other way, as read from the abstracts by an AI model (AI-written). Why studies differ.
- Experimental evidence on the productivity effects of generative artificial intelligence
Productivity: its finding supports the claim · Shakked Noy and Whitney Zhang · 2023 - The simple macroeconomics of AI
Productivity: its finding goes against the claim · Daron Acemoğlu · 2024
Related papers
On the same debate questions, key papers first; related by topic, not by method or conclusion. Browse all: Productivity · Jobs.
- How Might Fiscal Policy Respond to the Rise of Artificial Intelligence?
Karen Dynan et al. · NBER · 2026 - Forecasting the Economic Effects of AI
Ezra Karger et al. · NBER · 2026 - Task Expansion with Generative AI: Experimental Evidence on Middle-Skilled, Early-Career Workers
Hanna Brosch et al. · CESifo · 2026 - Large Language Models, Small Labor Market Effects
Anders Humlum and Emilie Vestergaard · NBER · 2025
Overview assembled automatically from collected metadata. How this site is built.