Echoes of AI : Investigating the downstream effects of AI assistants on software maintainability
Borg, Markus LU
; Hewett, Dave
; Hagatulah, Nadim
LU
; Couderc, Noric
LU
; Söderberg, Emma
LU
; Graham, Donald
; Kini, Uttam
and Farley, Dave
(2026)
In Empirical Software Engineering
31(6).
- Abstract
Context: AI assistants, like GitHub Copilot and Cursor, are transforming software engineering. While several studies highlight productivity improvements, their impact on maintainability requires further investigation. Objective: This study investigates whether co-development with AI assistants affects software maintainability, specifically how easily other developers can evolve the resulting source code. Method: We conducted a two-phase, preregistered controlled experiment involving 151 participants, 95% of whom were professional developers. In Phase 1, participants added a new feature to a Java web application, with or without AI assistance. In Phase 2, a randomized controlled trial, new participants evolved these solutions without AI... (More)
Context: AI assistants, like GitHub Copilot and Cursor, are transforming software engineering. While several studies highlight productivity improvements, their impact on maintainability requires further investigation. Objective: This study investigates whether co-development with AI assistants affects software maintainability, specifically how easily other developers can evolve the resulting source code. Method: We conducted a two-phase, preregistered controlled experiment involving 151 participants, 95% of whom were professional developers. In Phase 1, participants added a new feature to a Java web application, with or without AI assistance. In Phase 2, a randomized controlled trial, new participants evolved these solutions without AI assistance. Results: Phase 2 revealed no significant differences in subsequent evolution with respect to completion time or code quality. Bayesian analysis suggests that any speed or quality improvements from AI use were at most small and highly uncertain. Observational results from Phase 1 corroborate prior research: using an AI assistant yielded a 30.7% median reduction in completion time, and habitual AI users showed an estimated 55.9% speedup. Conclusions: Overall, we did not detect systematic maintainability advantages or disadvantages when other developers evolved code co-developed with AI assistants. Within the scope of our tasks and measures, we observed no consistent warning signs of degraded code-level maintainability. Future work should examine risks such as code bloat from excessive code generation and cognitive debt as developers offload more mental effort to assistants.
(Less)
- author
- Borg, Markus
LU
; Hewett, Dave
; Hagatulah, Nadim
LU
; Couderc, Noric
LU
; Söderberg, Emma
LU
; Graham, Donald
; Kini, Uttam
and Farley, Dave
- organization
- publishing date
- 2026-11
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Code quality, Controlled experiment, Maintainability, Programming with AI, Software engineering
- in
- Empirical Software Engineering
- volume
- 31
- issue
- 6
- article number
- 161
- publisher
- Springer
- external identifiers
-
- scopus:105041265690
- ISSN
- 1382-3256
- DOI
- 10.1007/s10664-026-10889-1
- language
- English
- LU publication?
- yes
- id
- 19e3544a-0e90-4974-9d18-ee2fe038c866
- date added to LUP
- 2026-08-24 09:19:44
- date last changed
- 2026-08-24 09:19:44
@article{19e3544a-0e90-4974-9d18-ee2fe038c866,
abstract = {{<p>Context: AI assistants, like GitHub Copilot and Cursor, are transforming software engineering. While several studies highlight productivity improvements, their impact on maintainability requires further investigation. Objective: This study investigates whether co-development with AI assistants affects software maintainability, specifically how easily other developers can evolve the resulting source code. Method: We conducted a two-phase, preregistered controlled experiment involving 151 participants, 95% of whom were professional developers. In Phase 1, participants added a new feature to a Java web application, with or without AI assistance. In Phase 2, a randomized controlled trial, new participants evolved these solutions without AI assistance. Results: Phase 2 revealed no significant differences in subsequent evolution with respect to completion time or code quality. Bayesian analysis suggests that any speed or quality improvements from AI use were at most small and highly uncertain. Observational results from Phase 1 corroborate prior research: using an AI assistant yielded a 30.7% median reduction in completion time, and habitual AI users showed an estimated 55.9% speedup. Conclusions: Overall, we did not detect systematic maintainability advantages or disadvantages when other developers evolved code co-developed with AI assistants. Within the scope of our tasks and measures, we observed no consistent warning signs of degraded code-level maintainability. Future work should examine risks such as code bloat from excessive code generation and cognitive debt as developers offload more mental effort to assistants.</p>}},
author = {{Borg, Markus and Hewett, Dave and Hagatulah, Nadim and Couderc, Noric and Söderberg, Emma and Graham, Donald and Kini, Uttam and Farley, Dave}},
issn = {{1382-3256}},
keywords = {{Code quality; Controlled experiment; Maintainability; Programming with AI; Software engineering}},
language = {{eng}},
number = {{6}},
publisher = {{Springer}},
series = {{Empirical Software Engineering}},
title = {{Echoes of AI : Investigating the downstream effects of AI assistants on software maintainability}},
url = {{http://dx.doi.org/10.1007/s10664-026-10889-1}},
doi = {{10.1007/s10664-026-10889-1}},
volume = {{31}},
year = {{2026}},
}