What is it about?
Although model-driven software engineering (MDSE) has proven effective in managing complex systems, its industrial adoption remains limited by the substantial maintenance overhead required for models and the specialised skills demanded of developers. Meanwhile, advances in artificial intelligence (AI), particularly generative and agentic AI, have shown great promise in automating code-related tasks such as comprehension, generation, and defect detection. These capabilities are largely powered by ’big code’: vast repositories of open-source software that now form the basis of data-driven, empirical SE and automated quality assurance. This paper aims to synthesise these two domains by exploring the integration of AI into model-driven practices. It provides a comprehensive overview of the current state of AI-augmented software engineering and introduces a novel taxonomy ’ai4se’ to classify and connect diverse AI applications within the field. On this basis, the paper proposes a vision for ’big models’ in software engineering (SE), an approach designed to leverage the structural advantages of MDSE alongside the scalability of AI. Finally, the paper discusses the pair modelling paradigm as a collaborative framework for the MDSE industry, designed to enhance software quality through human–AI partnership.
Featured Image
Photo by Aerps.com on Unsplash
Why is it important?
As software systems become increasingly complex and interconnected, traditional, code-centric development practices are struggling to keep pace with the demands for reliability, quality, and sustainability. High-profile system failures demonstrate that software complexity is no longer merely a technical challenge - it can have significant societal and economic consequences. Model-Driven Software Engineering (MDSE) provides powerful abstractions for managing this complexity, but its industrial adoption has been constrained by the substantial effort and specialized expertise required to create, maintain, and synchronize models with evolving code. This research shows how AI can help overcome this long-standing barrier by automating model creation, synchronization, and maintenance, making disciplined, model-centric engineering more practical at scale. The paper provides the foundation for this shift in three ways. First, the ai4se taxonomy and ontology establish a comprehensive framework for systematically connecting AI applications in software engineering across purpose, target, AI type, and autonomy level. Second, the Model Naturalness Hypothesis proposes that software models contain learnable statistical structures, opening new possibilities for using AI to analyse, generate, and reason about software architectures. Third, the proposed pair modelling paradigm describes how humans and AI can collaboratively evolve models and code, combining AI's ability to process and generate complex artefacts with human responsibility for intent, verification, accountability, and quality. Together, these contributions point towards a new form of AI-augmented software engineering: moving beyond AI that merely generates code towards AI that helps engineers understand, design, verify, and continuously evolve complex software systems.
Perspectives
This work grew out of my keynote at the 16th System Analysis and Modeling Conference (SAM 2024) at Johannes Kepler University in Linz, Austria, as part of the 27th International Conference on Model Driven Engineering Languages and Systems (MODELS 2024). What started as an attempt to make sense of the rapidly emerging field of AI for Software Engineering has since evolved into a broader research effort to understand how AI is changing the way we engineer software. The systematic literature review captures the state of the field up to Summer 2025. Since then, the field has continued to evolve rapidly, with important new developments emerging at the International Conference on Software Engineering (ICSE) and other venues. To me, this makes the taxonomy more valuable, not less: its core dimensions have proven remarkably robust, while its open structure allows it to evolve alongside the field. `ai4se` is deliberately designed as an open-source ontology, published under a CC-BY-SA licence. This means that the taxonomy does not have to remain a static snapshot in a paper. It can become a 'living knowledge resource' that the research community can continuously extend with new methods, tools, publications, and research directions. I would therefore be very happy to connect with researchers and practitioners interested in extending the `ai4se` ontology, incorporating emerging AI4SE research, and taking the systematic literature review forward together. I see this as an opportunity to turn a taxonomy into a shared community resource - and to help build a clearer and continuously evolving map of where AI is taking software engineering.
Ina Schieferdecker
Technische Universitat Berlin
Read the Original
This page is a summary of: Augmenting software engineering with AI. The ai4se taxonomy and its use, Innovations in Systems and Software Engineering, August 2026, Springer Science + Business Media,
DOI: 10.1007/s11334-026-00652-6.
You can read the full text:
Contributors
The following have contributed to this page
