Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Does AI Help or Hurt Learning?

Franco, Catalina ; Irmert, Natalie LU and Isaksson, Siri (2026) In Working Paper p.1-59
Abstract
AI is transforming how students learn, raising concerns about whether it expands educational opportunities or widens existing gaps. We examine this question in a preregistered lab experiment (N=572) in which students study a novel topic under one of three conditions: browsing only (control), AI-assisted, or AI-guided, and then complete an exam without AI access. We find no overall effect of AI access on learning outcomes. However, this average zero effect masks substantial heterogeneity. High-GPA women appear to benefit the most from AI-guided access, while the effects on men and low-GPA students are weaker and in some cases negative. We also find that students with AI access attempt fewer practice questions during the study phase. This... (More)
AI is transforming how students learn, raising concerns about whether it expands educational opportunities or widens existing gaps. We examine this question in a preregistered lab experiment (N=572) in which students study a novel topic under one of three conditions: browsing only (control), AI-assisted, or AI-guided, and then complete an exam without AI access. We find no overall effect of AI access on learning outcomes. However, this average zero effect masks substantial heterogeneity. High-GPA women appear to benefit the most from AI-guided access, while the effects on men and low-GPA students are weaker and in some cases negative. We also find that students with AI access attempt fewer practice questions during the study phase. This suggests that studying with AI crowds out other learning activities, but does not lead to an overall change in exam performance. Finally, exploratory analyses of prompt data provide suggestive evidence on why some students benefit more than others. More delegative AI use (measured by copy-pasting practice questions into the chatbot) is associated withattempting more questions but performing worse on the final exam. High-GPA women rely on this strategy the least and perform the best. Overall, AI appears to crowd out some traditional study effort without reducing learning on average, but because its benefits are concentrated among already advantaged students, it may reinforce existing educational inequalities. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Working paper/Preprint
publication status
published
subject
keywords
Generative AI, learning outcomes, study behavior, educational inequality, randomized experiment, human-AI interaction, C90, D83, I23, J16, O33
in
Working Paper
issue
2026:2
pages
1 - 59
language
English
LU publication?
yes
id
4c5d3458-4561-4e97-9411-cee346f6c984
date added to LUP
2026-04-09 15:04:11
date last changed
2026-04-09 15:04:11
@misc{4c5d3458-4561-4e97-9411-cee346f6c984,
  abstract     = {{AI is transforming how students learn, raising concerns about whether it expands educational opportunities or widens existing gaps. We examine this question in a preregistered lab experiment (N=572) in which students study a novel topic under one of three conditions: browsing only (control), AI-assisted, or AI-guided, and then complete an exam without AI access. We find no overall effect of AI access on learning outcomes. However, this average zero effect masks substantial heterogeneity. High-GPA women appear to benefit the most from AI-guided access, while the effects on men and low-GPA students are weaker and in some cases negative. We also find that students with AI access attempt fewer practice questions during the study phase. This suggests that studying with AI crowds out other learning activities, but does not lead to an overall change in exam performance. Finally, exploratory analyses of prompt data provide suggestive evidence on why some students benefit more than others. More delegative AI use (measured by copy-pasting practice questions into the chatbot) is associated withattempting more questions but performing worse on the final exam. High-GPA women rely on this strategy the least and perform the best. Overall, AI appears to crowd out some traditional study effort without reducing learning on average, but because its benefits are concentrated among already advantaged students, it may reinforce existing educational inequalities.}},
  author       = {{Franco, Catalina and Irmert, Natalie and Isaksson, Siri}},
  keywords     = {{Generative AI; learning outcomes; study behavior; educational inequality; randomized experiment; human-AI interaction; C90; D83; I23; J16; O33}},
  language     = {{eng}},
  note         = {{Working Paper}},
  number       = {{2026:2}},
  pages        = {{1--59}},
  series       = {{Working Paper}},
  title        = {{Does AI Help or Hurt Learning?}},
  url          = {{https://lup.lub.lu.se/search/files/246990414/WP26_2.pdf}},
  year         = {{2026}},
}