The Augmentation Trap: AI Productivity and the Cost of Cognitive Offloading

• URL: http://arxiv.org/abs/2604.03501v1
• Date: Fri, 03 Apr 2026 22:50:32 GMT
• Title: The Augmentation Trap: AI Productivity and the Cost of Cognitive Offloading
• Abstract summary: Experimental evidence confirms that AI tools raise worker productivity, but that sustained use can erode the expertise on which those gains depend.<n>We develop a dynamic model in which a decision-maker chooses AI usage intensity for a worker over time.<n>We decompose the tool's productivity effect into two channels, one independent of worker expertise and one that scales with it.
• Score: 0.0
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Experimental evidence confirms that AI tools raise worker productivity, but also that sustained use can erode the expertise on which those gains depend. We develop a dynamic model in which a decision-maker chooses AI usage intensity for a worker over time, trading immediate productivity against the erosion of worker skill. We decompose the tool's productivity effect into two channels, one independent of worker expertise and one that scales with it. The model produces three main results. First, even a decision-maker who fully anticipates skill erosion rationally adopts AI when front-loaded productivity gains outweigh long-run skill costs, producing steady-state loss: the worker ends up less productive than before adoption. Second, when managers are short-termist or worker skill has external value, the decision-maker's optimal policy turns steady-state loss into the augmentation trap, leaving the worker worse off than if AI had never been adopted. Third, when AI productivity depends less on worker expertise, workers can permanently diverge in skill: experienced workers realize their full potential while less experienced workers deskill to zero. Small differences in managerial incentives can determine which path a worker takes. The productivity decomposition classifies deployments into five regimes that separate beneficial adoption from harmful adoption and identifies which deployments are vulnerable to the trap.

Related papers

• The Boundaries of Automation: A Theory of Persistent Human Participation [107.84677954791773]
  We argue that human participation may persist even with highly capable AI systems.<n>In some activities, the target is not fully specified in advance but instead emerges through the interaction itself.<n>This perspective has important implications for the limits of automation.
  arXiv  Detail & Related papers  (2026-07-23T17:33:43Z)
• Managing the Human Fallback: Skill Investment Under Improving AI and Worker Mobility [0.0]
  We show that when firms deploy autonomous AI, they must decide how much work to leave to the system and how much to keep workers engaged.<n>We develop a parsimonious two-period model in which AI may outperform the worker when it functions, but may fail with positive probability.<n>A firm chooses worker engagement; engagement lowers current output for below-benchmark workers, but changes future skill through learning and erosion.
  arXiv  Detail & Related papers  (2026-06-27T23:34:23Z)
• Framing, Judging, Steering: An Assessable Competency Model for Teach-ing Students to Reason With Generative AI [47.027290803102666]
  We propose CoRe-3 (Co-Reasoning), a competency model factoring productive AI use into three assessable skills.<n>We ground the skills in theory, state five testable propositions, and instantiate them in CoReasoningLab.<n>Human-rater agreement and outcomes are next; we release the instrument, data, and protocol.
  arXiv  Detail & Related papers  (2026-06-04T10:25:31Z)
• Queue & AI: When Faster Tasks Slow Down the Workflow [2.89908994976586]
  The deployment of AI tools in customer service, writing, software development, and consulting operations has been reported to generate large per-task productivity gains.<n>We argue that such mean-based metrics can misrepresent AI's effects in where tasks accumulate and compete for scarce human attention.
  arXiv  Detail & Related papers  (2026-05-26T15:57:41Z)
• Short-Term Gain, Long-Term Fragility: AI Labor Substitution and the Erosion of Sustainable Capability [0.0]
  This paper develops a mechanism of capability masking and capability erosion under AI labor substitution.<n> AI-generated output can create the appearance that organizational capability has been replaced, even when dependence on skilled human labor remains.<n>Evidence from AI-assisted coding shows that generated output still requires substantial human verification and remains uneven in correctness, maintainability, and security.
  arXiv  Detail & Related papers  (2026-04-23T03:11:27Z)
• When Should a Robot Think? Resource-Aware Reasoning via Reinforcement Learning for Embodied Robotic Decision-Making [68.12864562049957]
  Embodied robotic systems increasingly rely on large language model (LLM)-based agents to support high-level reasoning.<n>This raises a fundamental question for embodied agents: when should the agent reason, and when should it act?<n>We propose RARRL (Resource-Aware Reasoning via Reinforcement Learning), a hierarchical framework for resource-aware orchestration of embodied agents.
  arXiv  Detail & Related papers  (2026-03-17T15:38:50Z)
• Barriers to AI Adoption: Image Concerns at Work [0.0]
  I find that workers adopt AI recommendations at lower rates when their reliance on AI is visible to the evaluator.<n>I introduce a novel incentive-compatible elicitation method showing that workers fear heavy reliance on AI signals a lack of confidence in their own judgment.
  arXiv  Detail & Related papers  (2025-11-23T18:50:34Z)
• AI Failure Loops in Devalued Work: The Confluence of Overconfidence in AI and Underconfidence in Worker Expertise [22.485163311963493]
  We examine the case of feminized labor: a class of devalued occupations historically misnomered as women's work''<n>We show how misjudgments on the automatability of workers' skills can lead to AI deployments that fail to bring value to workers.
  arXiv  Detail & Related papers  (2025-11-07T01:51:57Z)
• How Do AI Agents Do Human Work? Comparing AI and Human Workflows Across Diverse Occupations [112.57167042285437]
  We study how agents do human work by presenting the first direct comparison of human and agent workers.<n>We find that agents deliver results 88.3% faster and cost 90.4-96.2% less than humans.
  arXiv  Detail & Related papers  (2025-10-26T18:10:22Z)
• Training for Obsolescence? The AI-Driven Education Trap [0.0]
  We model an educational planner whose decision to adopt AI is driven by its teaching productivity, failing to internalize AI's future wage-suppressing effect on those same skills.<n>Our findings caution that policies promoting AI in education, if not paired with forward-looking labor market signals, may paradoxically undermine students' long-term human capital.
  arXiv  Detail & Related papers  (2025-08-27T07:04:19Z)
• Experimental Evidence That AI-Managed Workers Tolerate Lower Pay Without Demotivation [7.306174397662034]
  Experimental evidence on worker responses to AI management remains mixed, partly due to limitations in experimental fidelity.<n>We address these limitations with a customized workplace in the Minecraft platform, enabling high-resolution behavioral tracking of autonomous task execution.<n>Workers completed repeated production tasks under either human, AI, or hybrid management.<n>An AI manager trained on human-defined evaluation principles systematically assigned lower performance ratings and reduced wages by 40%, without adverse effects on worker motivation and sense of fairness.
  arXiv  Detail & Related papers  (2025-05-27T20:40:18Z)
• How Performance Pressure Influences AI-Assisted Decision Making [52.997197698288936]
  We show how pressure and explainable AI (XAI) techniques interact with AI advice-taking behavior.<n>Our results show complex interaction effects, with different combinations of pressure and XAI techniques either improving or worsening AI advice taking behavior.
  arXiv  Detail & Related papers  (2024-10-21T22:39:52Z)
• AI and Jobs: Has the Inflection Point Arrived? Evidence from an Online Labor Platform [0.13124513975412255]
  We investigate how AI influences freelancers across different online labor markets (OLMs) To shed light on the underlying mechanisms, we developed a Cournot-type competition model. We find that U.S. web developers tend to benefit more from the release of ChatGPT compared to their counterparts in other regions.
  arXiv  Detail & Related papers  (2023-12-07T10:06:34Z)
• Is the Most Accurate AI the Best Teammate? Optimizing AI for Teamwork [54.309495231017344]
  We argue that AI systems should be trained in a human-centered manner, directly optimized for team performance. We study this proposal for a specific type of human-AI teaming, where the human overseer chooses to either accept the AI recommendation or solve the task themselves. Our experiments with linear and non-linear models on real-world, high-stakes datasets show that the most accuracy AI may not lead to highest team performance.
  arXiv  Detail & Related papers  (2020-04-27T19:06:28Z)
• Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making [53.62514158534574]
  We study whether features that reveal case-specific model information can calibrate trust and improve the joint performance of the human and AI. We show that confidence score can help calibrate people's trust in an AI model, but trust calibration alone is not sufficient to improve AI-assisted decision making.
  arXiv  Detail & Related papers  (2020-01-07T15:33:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.