A Comparative Study of Human-Written Code and AI-Generated Code

AI has made it easier to generate code, explore possible solutions, and reduce some repetitive programming work. These tools can be helpful for students and experienced developers, but their growing use also raises practical questions about correctness, efficiency, security, and maintainability. This study compares human-written and AI-generated code to better understand where each approach may be useful and where caution is needed. The comparison considers correctness, execution time, memory use, readability, maintainability, security, and development effort. These factors matter because useful software must do more than produce the expected answer once: it should also be understandable, reasonably efficient, secure, and straightforward to update. Even when AI produces code quickly, testing and debugging may still be needed before the solution can be relied on. Human programmers bring reasoning, context, creativity, and an understanding of project-specific needs. AI coding tools can offer quick drafts, suggestions, and help with routine tasks. Neither approach should be assumed to be best in every case. The outcome depends on the problem, the implementation, the programmer’s experience, and how carefully the solution is tested. The main takeaway is that AI can be a useful assistant, but it should not replace careful programming judgment. Combining AI-generated suggestions with human review, testing, and verification may help developers work more effectively while still producing reliable, maintainable software. Any firm conclusion about comparative performance must be based on the benchmark results actually collected, with the study’s limitations clearly acknowledged.

1.                B. Yetiştiren, I. Özsoy, M. Ayerdem, and E. Tüzün, “Evaluating the Code Quality of AI-Assisted Code Generation Tools: An Empirical Study on GitHub Copilot, Amazon CodeWhisperer, and ChatGPT,” arXiv preprint arXiv:2304.10778, 2023. Available: https://arxiv.org/abs/2304.10778 2.                A. Moradi Dakhel, V. Majdinasab, A. Nikanjam, F. Khomh, M. C. Desmarais, and Z. M. J. Jiang, “GitHub Copilot AI pair programmer: Asset or Liability?” Journal of Systems and Software, vol. 203, article 111734, 2023. doi: 10.1016/j.jss.2023.111734. 3.                H. Pearce, B. Ahmad, B. Tan, B. Dolan-Gavitt, and R. Karri, “Asleep at the Keyboard? Assessing the Security of GitHub Copilot’s Code Contributions,” in Proceedings of the 2022 IEEE Symposium on Security and Privacy (SP), pp. 754–768, 2022. doi: 10.1109/SP46214.2022.9833571. 4.                C. Chen et al., “Evaluating Large Language Models Trained on Code,” arXiv preprint arXiv:2107.03374, 2021. Available: https://arxiv.org/abs/2107.03374 5.                J. Austin et al., “Program Synthesis with Large Language Models,” arXiv preprint arXiv:2108.07732, 2021. Available: https://arxiv.org/abs/2108.07732 V. Majdinasab et al., “Assessing the Security of GitHub Copilot Generated Code—A Targeted Replication Study,” arXiv preprint arXiv:2311.11177, 2023. Available: https://arxiv.org/abs/2311.11177