A research paper published on arXiv traces the evolution of artificial intelligence in systems engineering (SE) over the past decade, identifying five critical research gaps and releasing an interactive web tool for practitioners. The study, authored by Bank, H Sinan, Herber, Daniel R, and Bradley, Thomas, is titled "AI4SE and SE4AI Exploration: A Decade Looking Back and Forward" and examines the intersection of AI and systems engineering from both directions: using AI to improve SE (AI4SE) and adapting SE practices to develop AI systems (SE4AI).
The paper notes that the March 2020 INCOSE INSIGHT special issue on AI and Systems Engineering became the most downloaded issue in the publication's history. That special issue launched a research community that now draws over 250 registrants to its annual workshop.
The authors trace progress across three phases:
| Phase | Label | Description (from source) |
|---|---|---|
| 1 | Foundational | Early work establishing AI/SE concepts |
| 2 | Applied | Practical implementations and case studies |
| 3 | LLM inflection | Recent impact of large language models on SE |
To assess the field's evolution, the researchers performed a human-AI agreement literature review using both human expertise and six AI models to evaluate the relevance of 1,712 INCOSE INSIGHT articles and 889 SERC publications. The results identified five critical research gaps (not enumerated in the source) that the community has not yet closed. The paper offers guidance for practitioners navigating AI adoption, assurance, and workforce transformation in SE.
The authors also released the AI4SE/SE4AI Explorer web application so that readers can compare their own relevance judgments with the human and AI raters. The agreement data from the review is also shared.
For enterprise technology leaders, the research highlights the need for structured approaches to integrating AI into existing engineering workflows. The three-phase framework (foundational, applied, LLM inflection) provides a lens for evaluating their organization's maturity. The identified gaps are likely to inform future research priorities and tool development. The availability of the AI4SE/SE4AI Explorer allows practitioners to benchmark their own perspectives against the community's collective assessment.
