Generative AI is no longer just an emerging topic in software engineering education. It is already changing how software engineering is taught, how students learn, and how their competencies are assessed.
In a recent survey, iSAQB asked Academic Partners and educators about their experiences with generative AI in teaching and curriculum development. Based on 15 responses, the survey provides an exploratory snapshot of a field undergoing rapid change.
One message stands out: AI is not simply changing the tools software engineers use. It is changing what software engineers need to be good at.
AI Has Already Arrived in Software Engineering Education
Almost all respondents report changes in their teaching as a result of generative AI. More than half describe at least moderate integration of AI into their courses, while others report significant redesigns or even fundamental transformations.
AI is also becoming part of everyday teaching practice. Educators are integrating it into assignments and exercises, experimenting with dedicated AI content, and adapting course design accordingly.
However, greater efficiency comes with a potential downside. More than half of respondents observe that students are becoming more dependent on AI tools and paying less attention to fundamentals. A similar proportion report greater efficiency accompanied by reduced depth of understanding.
This creates a fundamental challenge for software engineering education: Students can increasingly produce convincing results without necessarily understanding how or why those results work.
Assessment Has to Evolve
This development is already reflected in how educators assess competence. 73% of respondents use assignments with explicit guidelines for AI usage, while 60% use oral examinations or live coding sessions. Project-based assessments, process-oriented evaluation, and AI-aware assessment formats are also being used.
The implication goes beyond examinations. When AI can generate code, documentation, tests, and even architectural proposals, evaluating the final artifact alone becomes less meaningful. Educators increasingly need to make students’ understanding, reasoning, and decision-making visible.
The focus is shifting from “What did you produce?” toward “Why is this the right solution?”
The Biggest AI Competency Gap May Not Be Prompting, but Judgment
One of the most striking findings concerns competency gaps. When asked which GenAI-related competencies students and junior software engineers currently lack most, 80% identified the ability to critically evaluate and validate AI-generated results. Another 67% highlighted the ability to integrate AI tools into structured development workflows. Prompt engineering, by comparison, was selected by only 13%.
The message is significant: interacting effectively with an AI system matters, but engineering judgment appears to matter considerably more. Future software professionals need to determine whether an AI-generated solution is correct, secure, maintainable, appropriate for its context, and compatible with the architecture of the overall system.
Generating a solution is becoming easier. Judging whether it is a good solution remains an engineering responsibility.
AI May Transform Roles Rather Than Simply Eliminate Them
The respondents also paint a more nuanced picture of how AI may affect software engineering careers. When discussing AI-related career concerns with students, 80% of respondents emphasize that new roles and competencies will emerge, while 67% emphasize that AI is more likely to augment rather than replace developers.
This suggests that educators tend to frame AI primarily as a transformation of software engineering roles rather than their simple elimination. The likely transformation is therefore not simply from “human developer” to “AI developer.” Instead, the role itself may move toward greater emphasis on decision-making, validation, integration, architecture, and responsibility.
Why Software Architecture May Become Even More Important
This development has particular implications for software architecture. AI can generate solutions quickly, but generating a solution is not the same as making a sound architectural decision. Architects still need to understand context, constraints, dependencies, quality requirements, risks, and trade-offs.
As AI-generated components and AI-enabled systems become more common, these capabilities may become even more important. This view is reflected in the survey: 80% of respondents identify the development of best practices for AI in software architecture as an important priority.
The question is therefore increasingly not whether software architects will use AI, but how they can use it without surrendering architectural judgment.
Responsibility Becomes an Engineering Competency
The survey also reveals strong concern about the ethical and professional implications of AI. Respondents highlighted accountability and responsibility, transparency and explainability, and security and misuse risks in particular. Privacy and bias were also identified as relevant concerns.
For software architects, this matters because architectural decisions can determine how AI systems behave, how transparent they are, what data they use, which dependencies they introduce, and how risks can be controlled.
Responsible AI therefore cannot be treated exclusively as a legal or compliance issue. It is increasingly becoming part of engineering itself.
What This Means for Software Engineering Education
Taken together, the findings suggest a broader transition. Generative AI can substantially increase the productivity of software professionals, but productivity alone does not create good software.
As producing software artifacts becomes easier, the differentiating capabilities of software engineers and architects may increasingly be their ability to:
- critically evaluate and validate AI-generated results;
- understand systems and their underlying principles;
- make sound architectural decisions;
- integrate AI into structured engineering processes;
- identify limitations, dependencies, and risks;
- take responsibility for the systems they build.
Software engineering education will need to evolve accordingly. For iSAQB, this means continuing to explore how AI affects software architecture, which competencies architects will need in an AI-driven environment, and how these competencies should be reflected in education and professional practice.
AI may change how we build software. The ability to make good architectural decisions remains essential.
About the Survey
The survey was conducted among iSAQB Academic Partners and educators involved in software engineering teaching and curriculum development.
It received 15 responses and should be understood as an exploratory snapshot rather than a statistically representative study. The findings reflect educators’ experiences, perceptions, and expectations rather than objective measurements of educational outcomes.
Nevertheless, the responses reveal clear patterns and provide a basis for further discussion about the future of software engineering and software architecture education.
About the Survey Compilation
The survey results were compiled by Dr.-Ing. Meng Zhang, a postdoctoral researcher with a Ph.D. in Mechatronics, Robotics, and Automation Engineering from Clausthal University of Technology (TU Clausthal). His expertise includes machine learning and software development. Meng is also an iSAQB member and accredited CPSA-Foundation Level trainer.