Saved in:
Bibliographic Details
Main Authors: Belotti, Federico, Coniglio, Stefano, Cosma, Antonio, Fallucchi, Francesco
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.23920
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917526026321920
author Belotti, Federico
Coniglio, Stefano
Cosma, Antonio
Fallucchi, Francesco
author_facet Belotti, Federico
Coniglio, Stefano
Cosma, Antonio
Fallucchi, Francesco
contents Real-effort tasks, in which participants perform cognitively costly activities whose outcomes depend on actual performance, are widely used in experimental economics. Their validity, however, rests on the assumption that a human performs them. We study whether this assumption still holds in the era of Artificial Intelligence (AI) and Large Language Models (LLMs). Using 8 canonical real-effort tasks and 23 LLMs from three major providers, we show that most tasks can now be solved accurately and at a negligible cost, while only a few resist automation. Performance improves with each model generation, and midtier models are rapidly closing the gap with frontier ones, broadening the set of widely accessible models that can automate these tasks. Additionally, we show that verbally offering monetary incentives has no effect on LLM performance. Our findings establish a boundary condition for the use of real-effort tasks in unsupervised settings: when participants can cheaply outsource task completion to an LLM, observed performance may no longer reflect genuine human effort.
format Preprint
id arxiv_https___arxiv_org_abs_2605_23920
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Artificial Effort
Belotti, Federico
Coniglio, Stefano
Cosma, Antonio
Fallucchi, Francesco
Computers and Society
Artificial Intelligence
Real-effort tasks, in which participants perform cognitively costly activities whose outcomes depend on actual performance, are widely used in experimental economics. Their validity, however, rests on the assumption that a human performs them. We study whether this assumption still holds in the era of Artificial Intelligence (AI) and Large Language Models (LLMs). Using 8 canonical real-effort tasks and 23 LLMs from three major providers, we show that most tasks can now be solved accurately and at a negligible cost, while only a few resist automation. Performance improves with each model generation, and midtier models are rapidly closing the gap with frontier ones, broadening the set of widely accessible models that can automate these tasks. Additionally, we show that verbally offering monetary incentives has no effect on LLM performance. Our findings establish a boundary condition for the use of real-effort tasks in unsupervised settings: when participants can cheaply outsource task completion to an LLM, observed performance may no longer reflect genuine human effort.
title Artificial Effort
topic Computers and Society
Artificial Intelligence
url https://arxiv.org/abs/2605.23920