Will It Survive? Deciphering the Fate of AI-Generated Code in Open Source

The integration of AI agents as coding assistants into software development has raised questions about the long-term viability of AI agent-generated code. A prevailing hypothesis within the software engineering community suggests this code is"disposable", meaning it is merged quickly but discarded shortly thereafter. If true, organizations risk shifting maintenance burden from generation to post-deployment remediation. We investigate this hypothesis through survival analysis of 201 open-source projects, tracking over 200,000 code units authored by AI agents versus humans. Contrary to the disposable code narrative, agent-authored code survives significantly longer: at the line level, it exhibits a 15.8 percentage-point lower modification rate and 16% lower hazard of modification (HR = 0.842, p<0.001). However, modification profiles differ. Agent-authored code shows modestly elevated corrective rates (26.3% vs. 23.0%), while human code shows higher adaptive rates. However, the effect sizes are small (Cram\'er's V = 0.116), and per-agent variation exceeds the agent-human gap. Turning to prediction, textual features can identify modification-prone code (AUC-ROC = 0.671), but predicting when modifications occur remains challenging (Macro F1 = 0.285), suggesting timing depends on external organizational dynamics. The bottleneck for agent-generated code may not be generation quality, but the organizational practices that govern its long-term evolution.

• Publication year

  2026

• Venue

  arXiv.org

• Publication date

  2026-01-23

• Fields of study

  Computer Science

• Identifiers
  DOI 10.48550/arXiv.2601.16809arXiv 2601.16809
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Where Do AI Coding Agents Fail? An Empirical Study of Failed Agentic Pull Requests in GitHub

  2026cited by this paper
• The Rise of AI Teammates in Software Engineering (SE) 3.0: How Autonomous Coding Agents Are Reshaping Software Engineering

  2025cited by this paper
• A First Look at Conventional Commits Classification

  2025cited by this paper
• DVC in Open Source ML-development: The Action and the Reaction

  2024cited by this paper
• Understanding the Time to First Response in GitHub Pull Requests

  2023cited by this paper
• SWE-bench: Can Language Models Resolve Real-World GitHub Issues?

  2023cited by this paper
• SPI: Automated Identification of Security Patches via Commits

  2021cited by this paper
• Asleep at the Keyboard? Assessing the Security of GitHub Copilot’s Code Contributions

  2021cited by this paper
• Evaluating Large Language Models Trained on Code

  2021cited by this paper
• PyExplainer: Explaining the Predictions of Just-In-Time Defect Models

  2021cited by this paper
• CodeBLEU: a Method for Automatic Evaluation of Code Synthesis

  2020cited by this paper
• Explainable AI for Software Engineering

  2020cited by this paper
• Natural Software Revisited

  2019cited by this paper
• A comprehensive study on deep learning bug characteristics

  2019cited by this paper
• Taxonomy of Real Faults in Deep Learning Systems

  2019cited by this paper
• A Survival Analysis-Based Prioritization of Code Checker Warning: A Case Study Using PMD

  2019cited by this paper
• AutoSpearman: Automatically Mitigating Correlated Software Metrics for Interpreting Defect Models

  2018cited by this paper
• Boosting Automatic Commit Classification Into Maintenance Activities By Utilizing Source Code Changes

  2017cited by this paper
• Curating GitHub for engineered software projects

  2017cited by this paper
• An Empirical Comparison of Model Validation Techniques for Defect Prediction Models

  2017cited by this paper
• Developer Turnover in Global, Industrial Open Source Projects: Insights from Applying Survival Analysis

  2017cited by this paper
• Modeling Survival Data Extending The Cox Model

  2016cited by this paper
• “Why Should I Trust You?”: Explaining the Predictions of Any Classifier

  2016cited by this paper
• Your Chi-Square Test Is Statistically Significant: Now What?.

  2015cited by this paper
• Code Ownership and Software Quality: A Replication Study

  2015cited by this paper
• Predicting bug-fixing time: An empirical study of commercial software projects

  2013cited by this paper
• Scikit-learn: Machine Learning in Python

  2011cited by this paper
• Don't touch my code!: examining the effects of ownership on software quality

  2011cited by this paper
• Automatic classication of large changes into maintenance categories

  2009cited by this paper
• Applied Regression Analysis and Generalized Linear Models

  2008cited by this paper
• SMOTE: Synthetic Minority Over-sampling Technique

  2002cited by this paper
• Predicting Fault Incidence Using Software Change History

  2000cited by this paper
• Identifying reasons for software changes using historic databases

  2000cited by this paper
• Survival Analysis: A Self-Learning Text

  1997cited by this paper
• An introduction to categorical data analysis

  1997cited by this paper
• Software Engineering Economics

  1993cited by this paper
• Partial residuals for the proportional hazards regression model

  1982cited by this paper
• The dimensions of maintenance

  1976cited by this paper
• Evaluation of survival data and two new rank order statistics arising in its consideration.

  1966cited by this paper
• Nonparametric Estimation from Incomplete Observations

  1958cited by this paper
• Regression Models and Life-Tables

  year unknowninfluential reference

• No citing papers are available for this paper.