MétaCan
Menu
← Back to cohort

Different by Design: Understanding Human-AI Collaboration Through GitHub Pull Requests

2025· article· W7125591383 on OpenAlexaff
Karla Gonzalez, Mariam El Mezouar

Bibliographic record

Venuenot available
Typearticle
Language
FieldComputer Science
TopicSoftware Engineering Research
Canadian institutionsRoyal Military College of Canada
Fundersnot available
KeywordsMaintainabilityCoding (social sciences)SoftwareBridge (graph theory)Cursor (databases)Descriptive statisticsStatistical analysisPresentation (obstetrics)

Abstract

fetched live from OpenAlex

Autonomous Coding Agents (ACAs) such as GitHub Copilot, Claude Code, and Cursor are transitioning from assistive back-pocket tools to active contributor status in collaborative software engineering projects. Their growing foothold raises questions about generated code quality, pull request (PR) acceptance rates and influencing factors, and long-term project impact. Building on the AIDev dataset, we conduct an investigation of how AI-authored PRs compare against human-authored PRs. We take the original descriptive study a step further by augmenting the AIDev dataset, deriving additional structural and linguistic metrics, and performing statistical analyses. We report statistically significant differences in rejection rates across distinct agent types and outline efforts to manually label failure modes in rejected AI PR and to track different maintainability impacts. This project aims to bridge descriptive characterization with predictive and diagnostic insight, laying yet another layer on the foundation for advancing tomorrow's ACAs.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.005
metaresearch head score (Gemma)0.039
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: Observational
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.007
Threshold uncertainty score0.028

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0050.039
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0030.002
Science and technology studies0.0010.001
Scholarly communication0.0040.005
Open science0.0010.003
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0020.001

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.073
GPT teacher head0.351
Teacher spread0.278 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same topicSoftware Engineering Research→French-language works237,207→