Newest papers

Documentation Drift in AI-Generated Pull Requests

Will De Michele, Kla Tantithamthavorn · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.27381 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Agen pengkodean AI sekarang berkontribusi pada repositori perangkat lunak pada skala, namun praktik dokumentasi mereka tetap tidak ditandai. Kami menganalisis 29.669 permintaan menarik yang dihasilkan oleh AI dari 2.585 repositori di dataset AIDev, mengukur pertukaran dokumen: apakah modifikasi file kode disertai dengan pembaruan file dokumentasi di jendela 90 hari yang sama. Kami menemukan bahwa 63.2% dari PR AI mengubah kode tanpa memperbarui dokumen apa pun. Ini bervariasi secara signifikan berdasarkan jenis tugas: 55.2% untuk fitur dan 77.5% untuk refactoring. File kode yang tidak didokumentasikan diikuti oleh perbaikan ke file yang sama 8.6 hingga 11.7% lebih sering daripada file yang didokumentasikan di jendela 30 hingga 90 hari. Perbandingan tingkat PR mentah menunjukkan sebaliknya karena PR yang didokumentasikan 3.9 kali lebih besar dan karena itu menarik lebih

Machine translation (Meta M2M100); the English original follows.

Abstract

AI coding agents now contribute to software repositories at scale, yet their documentation practices remain uncharacterised. We analyse 29,669 AI-generated pull requests from 2,585 repositories in the AIDev dataset, measuring documentation co-change: whether code-file modifications are accompanied by a documentation-file update in the same PR. We find that 63.2% of AI PRs change code without updating any documentation. This varies sharply by task type: 55.2% for features and 77.5% for refactoring. Undocumented code files are followed by a fix to the same file 8.6 to 11.7% more often than documented files across 30 to 90-day windows. A raw PR-level comparison shows the opposite because documented PRs are 3.9 times larger and therefore attract more fixes. Controlling for size exposure reverses this result. Documented changes also merge faster despite their larger size (median: 40 seconds versus 184 seconds), although merging is largely automated in this dataset. Documentation drift also compounds: within 90 days, a file changed without documentation is later re-documented at roughly half the rate of a file changed with documentation (24% versus 48%). This suggests that documentation debt tends to accumulate rather than be revisited.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Documentation, Merge (version control), Computer science, Internal documentation, Task (project management), Coding (social sciences)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange