First, the researchers classify the manifestations of GenAI in software engineering along two central dimensions:
Focus of augmentation (process vs. product):
This distinguishes whether AI primarily supports and optimizes traditional software engineering processes, or whether it becomes part of the software product itself by replacing traditionally programmed functions with integrated AI capabilities.
Degree of autonomy (passive vs. active):
Passive AI responds exclusively to explicit user input. Active or agentic AI systems, by contrast, have their own thread of control, act proactively, and make partially autonomous decisions within a system.
The combination of these two dimensions results in four fundamental forms of GenAI augmentation, as shown in the figure.
To better understand the impact of these four forms, the researchers used Marshall McLuhan’s tetrad analysis tool. Specifically, this involves answering the following questions: What is enhanced in software engineering by GenAI? What is reversed or turned into its opposite? What is retrieved from the past? What becomes obsolete?
Based on this analysis, the authors developed a roadmap that brings together key challenges and opportunities in the field of GenAI-augmented software engineering. These include, among others:
- advanced prompt engineering
- questions of accountability and security
- collaboration in hybrid human-AI teams
- coordination of process agents and product agents
In the paper, the researchers also formulate ten strategic predictions for the future of software engineering.
The article has been accepted for publication in the journal ACM Transactions on Software Engineering and Methodology (TOSEM). A preview is available here: https://doi.org/10.1145/3788879
Reference:
Domenico Amalfitano, Andreas Metzger, Marco Autili, Tommaso Fulcini, Tobias Hey, Jan Keim, Patrizio Pelliccione, Vincenzo Scotti, Anne Koziolek, Raffaela Mirandola, and Andreas Vogelsang. 2026. A Research Roadmap for Augmenting Software Engineering Processes and Software Products with Generative AI. ACM Trans. Softw. Eng. Methodol. Just Accepted (January 2026).
On July 7, 2026, the article will be presented as a Journal-First paper at FSE 2026 (ACM International Conference on the Foundations of Software Engineering (FSE) in Motreal. Link to the track: FSE 2026 - Journal-First Paper - FSE 2026