AI-Assisted Programming in Practice Conceptual Foundations and Data-Informed Observations
Keywords:
AI-Assisted Programming, Large Language Models, Task Specification, Interpretive Differences, Program Synthesis, Human-in-the-LoopAbstract
The increasing use of large language models in programming practice has transformed how software is designed, implemented, and validated. While existing studies largely focus on performance, productivity, or tool ecosystems, less attention has been paid to the role of requirement specification and interpretation in AI-assisted programming. This study examines how differences in task specification influence the behavior of humans and artificial intelligence during program construction.
The analysis is based on a comparative, practice-informed observation of programming tasks executed under two contrasting conditions: an incompletely specified task and a fully, explicitly specified task. Using a controlled programming context and the same execution environment, the study investigates how algorithmic assumptions, interpretative choices, and code structure emerge in each scenario for both human developers and AI-generated solutions. The focus is placed not on execution speed or optimality, but on interpretative variability, stability across executions, and qualitative properties of the resulting program code, including structural complexity and readability.
The observations show that incompletely specified tasks create a wide interpretative space in which AI-generated solutions exhibit substantial variability and, in some cases, exceed the intended scope of the task through functionally abundant or overly complex implementations. In contrast, explicit specification significantly reduces interpretative divergence and leads to conceptual convergence between human-written and AI-generated code. However, even under fully specified conditions, differences remain in qualitative aspects of code structure, with AI-generated solutions tending toward higher structural complexity.
These findings highlight specification as a critical boundary condition in AI-assisted programming and emphasize the evolving role of humans from code production toward interpretation, validation, and refinement in modern programming workflows.
Downloads
References
Chen, M., Tworek, J., Jun, H., Yuan, Q., Ponde de Oliveira Pinto, H., Kaplan, J., Edwards, H., Burda, Y., Joseph, N., Brockman, G., Ray, A., Puri, R., Krueger, G., Petrov, M., Khlaaf, H., Sastry, G., Mishkin, P., Chan, B., Gray, S., Ryder, N., … Zaremba, W. (2021). Evaluating large language models trained on code. arXiv. https://doi.org/10.48550/arXiv.2107.03374
Fan, G., Liu, D., Zhang, R., & Pan, L. (2025). The impact of AI-assisted pair programming on student motivation, programming anxiety, collaborative learning, and programming performance: a comparative study with traditional pair programming and individual approaches. International Journal of STEM Education, 12(1), 16.
Gottlander, J., & Khademi, T. (2023). The Effects of AI Assisted Programming in Software Engineering.
Li, Y., Choi, D., Chung, J., Kushman, N., Schrittwieser, J., Leblond, R., ... & Vinyals, O. (2022). Competition-level code generation with alphacode. Science, 378(6624), 1092-1097.
Nijkamp, E., Pang, B., Hayashi, H., Tu, L., Wang, H., Zhou, Y., ... & Xiong, C. (2022). Codegen: An open large language model for code with multi-turn program synthesis. arXiv preprint arXiv:2203.13474.
Ramasamy, D. (2024). Towards AI-Assisted Data Science Development: Decoding, Visualising, and Enhancing Human-AI Collaboration for Data Science Workflows in Practice (Doctoral dissertation, University of Zurich).
Ray, P. P. (2025). A Review on Vibe Coding: Fundamentals, State-of-the-art, Challenges and Future Directions. Authorea Preprints.
Tabatadze, B. (2024). Prospects of Using AI Technologies in Open Journal Systems (OJS). Journal of Technical Science and Technologies, 8(2), 68-74.
Tabatadze, B. (2025). Solving partial differential equations with deep neural networks. In Proceedings of the International Conference on Global Practice of Multidisciplinary Scientific Studies II. 763.
Taulli, T. (2024). AI-assisted programming: Better planning, coding, testing, and deployment. O'Reilly Media, Inc..
Wang, Y., Yu, P. D., & Tan, C. W. (2025). Future-Proofing Programmers: Optimal Knowledge Tracing for AI-Assisted Personalized Education. arXiv preprint arXiv:2509.23996.
