ARTFEED — Contemporary Art Intelligence

AI Assistance May Degrade Productivity, New Model Shows

ai-technology · 2026-05-13

A new model from arXiv (2605.11350) reveals that increased AI assistance can paradoxically reduce productivity. Researchers modeled human agents with varying skill levels exerting utility-maximizing effort with AI help. They found that when skill development or AI unreliability is endogenous, a productivity paradox emerges: more AI aid degrades output. Long-term, AI literacy gaps lead to skill polarization in steady state. The study highlights that AI's impact is not uniformly positive and depends on human adaptation.

Key facts

  • arXiv paper 2605.11350 proposes a model of human-AI interaction
  • Endogenous skill development or AI unreliability can induce a productivity paradox
  • Increased AI assistance may degrade productivity
  • Skill polarization emerges when accounting for heterogeneity in AI literacy
  • AI literacy is the capability to identify and adapt to inaccurate AI outputs
  • The model examines long-term distributional effects of AI on skill
  • Human agents with varying skill levels exert utility-maximizing effort
  • The study elucidates mechanisms by which AI affects productivity

Entities

Institutions

  • arXiv

Sources