Preprint IEEE ICSME 2026 To appear

CMM-AISE: A Cognitive Maturity Model for AI-Assisted Software Engineering

Daniel Alejandro González-Rueda1, Mario Linares-Vásquez1, Daniel Soto2, Andrés Hurtado3

1 — Universidad de los Andes, Bogotá, Colombia
2 — Tres Astronautas, Bogotá, Colombia
3 — Blend 360, Bogotá, Colombia

IEEE International Conference on Software Maintenance and Evolution (ICSME) 2026 · Industry / Vision Track · To appear

This is the preprint (author) version. It has been accepted for publication at ICSME 2026 but has not yet been through final copyediting or typesetting. The definitive version will appear in IEEE Xplore — please cite the published version once it is available.

Abstract

The rapid integration of artificial intelligence (AI) systems into software engineering (SE) is significantly transforming how researchers and practitioners perform tasks across the Software Development Lifecycle (SDLC). While prior work has largely focused on productivity gains and automation, significantly less attention has been given to how AI reshapes the cognitive development of software engineers. Traditional models of expertise assume a progression from novice to expert through increasingly complex reasoning, abstraction, and system-level understanding. However, AI systems can both support and disrupt this progression by redistributing cognitive effort between human and machine. In this paper, we argue that AI-assisted engineering introduces a critical tension between cognitive scaffolding, which enhances learning and reasoning, and cognitive substitution, which risks bypassing essential cognitive processes. Furthermore, we argue that substitution risk is not uniform across a practitioner’s skillset but peaks on the least developed skills where the judgment is less consolidated and harder to catch. To address this gap, we propose a Cognitive Maturity Model for AI-Assisted Software Engineering (CMM-AISE) that characterizes how engineers develop and exercise cognitive capabilities across SDLC stages under different levels of AI mediation. The model integrates capability dimensions, maturity levels, stage-dependent cognitive demands, and human responsibilities. We present how cognitive maturity manifests differently across SDLC stages and discuss implications for engineering practice, tool design, and education. Our work provides a foundation for understanding and guiding the future of human–AI collaboration in software engineering.

Index Terms
  • cognitive maturity
  • AI-assisted software engineering
  • SDLC
  • human–AI teaming
  • expertise development
How to cite
D.A. González-Rueda, M. Linares-Vásquez, D. Soto, and A. Hurtado, “CMM-AISE: A Cognitive Maturity Model for AI-Assisted Software Engineering,” Proc. IEEE International Conference on Software Maintenance and Evolution (ICSME), 2026, to appear.
BibTeX
@inproceedings{gonzalezrueda2026cmmaise,
  author    = {Gonz\'{a}lez-Rueda, Daniel Alejandro and
               Linares-V\'{a}squez, Mario and
               Soto, Daniel and
               Hurtado, Andr\'{e}s},
  title     = {{CMM-AISE}: A Cognitive Maturity Model for
               {AI}-Assisted Software Engineering},
  booktitle = {Proc. IEEE International Conference on Software
               Maintenance and Evolution (ICSME)},
  year      = {2026},
  note      = {To appear}
}

More papers on the way

Further work from Tres Astronautas, Universidad de los Andes, and Blend·Colombia — including research on specification graphs — will be published here.