Exploring the Scope of Software Engineering for AI: Insights from IEEE TSE

Artificial Intelligence (AI) and Machine Learning (ML) have revolutionized many fields, including software engineering. With the growing influx of AI-related research in Software Engineering (SE) conferences and journals, it has become increasingly important to clarify which contributions are most relevant for SE venues like IEEE Transactions on Software Engineering (TSE).

A recent editorial, written by the TSE editorial team, aims to offer guidance on scoping SE for AI. The editorial addresses key challenges in determining which AI-focused research truly advances the field of SE versus research better suited for AI or ML-specific venues. The editorial encourages contributions that extend beyond improving AI algorithms or models and directly tackle SE problems. Such contributions include, for example, testing, verifying, debugging, and repairing ML components within broader software systems or studying socio-technical aspects of AI integration in development teams.

The inception and writing of the editorial required several discussions and iterations. The idea originated from a weekly discussion by us (editor-chief and associate editors-in-chief), reflecting upon (i) the increasing number of submissions the journal is receiving on the topic, and (ii) at the same time, the non-negligible number of AI-related submissions that, unfortunately, turned out not to fit the journal scope, and (iii) the increasing number of nuanced discussions with associate editors on papers for which a scope decision was initially unclear.

The process started from the simple scope-checking guidelines that the TSE editorial team previously wrote for the editorial board, which were largely inspired by the guidelines some recent software engineering conferences (above all, ICSE) used. We collected insights from the journal’s associate editors in multiple meetings with them, in particular, an in-person meeting held in Lisbon during ICSE 2024 and two follow-up online meetings. Based on the brainstorming we had with associate editors, we wrote a first draft. Then, we iteratively refined it by

• Having dedicated online discussions with several editorial boards;
• Asking for feedback from some renowned research community members with great expertise on the topic. These are Matt Dwyer (North Carolina State University), Sebastian Elbaum (North Carolina State University), and Mark Harman (Meta and University College of London). We are truly grateful to them for the feedback they provided;
• Asking all editorial board members to comment and actively contribute to the draft; and
• Several meetings were held in which the three of us participated to refine and finalize the editorial.

This editorial serves as a starting point for a community-wide conversation about the boundaries between SE and AI, emphasizing the need for high-quality, relevant SE contributions. Whether you’re working on AI for SE or SE for AI, this is essential reading to understand how your research fits into the broader SE landscape.

Check out the full editorial in IEEE Transactions on Software Engineering to learn more and join the conversation!

Sebastian Uchitel

IEEE TSE editor-in-chief

Marsha Chechik and Massimiliano Di Penta

IEEE TSE associate editors-in-chief

Disclaimer: The posts in the SIGSOFT Blog are written by individual contributors and any views or opinions represented in their posts are personal, belong solely to the blog authors, and do not necessarily represent those of ACM SIGSOFT or ACM.