Artificial intelligence dominated the conversations at SEFI 2025, but does the literature reflect the same level of disruption? In this evidence-based blog, Professor Andrew Garrard explores how AI is currently appearing in engineering education research, revealing not only the scale of interest but the surprising breadth — and fragmentation — of themes emerging across the sector.
In September 2025, I attended the SEFI Annual
Conference in Tampere, Finland — arguably the largest and most important
engineering education conference in Europe. Like most conferences, the de facto
themes are defined not by the event organisers but by the delegates'
conversations during coffee and lunch breaks. Based on the conversations I had,
you could be forgiven for thinking the conference theme was artificial
intelligence. People weren't just talking about AI; there were meta-discussions
about the fact that everyone was talking about AI.
Inspired by xkcd.com
Alongside Mikko Nurminen from Tampere University, I
co-chair the SEFI Digital Learning Special Interest Group (SIG). During the
conference we held our annual in-person meeting and agreed that the SIG would
serve as SEFI's primary hub for AI-related research. This is similar to the
approach we've taken in the Engineering faculty at Sheffield, and I suspect it
reflects how universities are responding to AI more broadly. I have
reservations about it.
I consider myself an AI pragmatist. I'm neither
evangelical about the technology nor morally opposed to it (I'm reliably
informed by Gen Z that the latter position is called being an "AI
vegan"). I'm a voracious AI user, with personal paid subscriptions to
Anthropic and OpenAI, and I am routinely left forlorn when Claude informs me
I'm out of tokens. My justification is straightforward: AI supercharges my
productivity, and failing to be proficient with it risks making me obsolete.
There is no part of my job that AI doesn't touch. I use
it to proofread emails, build software to automate administrative tasks,
workshop teaching material and assessment design, and craft richer feedback for
students. It is also reshaping what I teach and how I assess — the existence of
an all-knowing artificial intelligence brings into question whether students
need to hold specific engineering knowledge at all. That context is precisely
why I'm sceptical about groups that focus on AI in education: the scope is the
problem.
Special interest groups exist so people with common
interests can exchange ideas and best practice. But when AI is used in
assessment, project-based learning, ethics, curriculum design, and more, there
is a real risk that a group formed around the tool itself contains people with
almost no meaningful overlap in how they use it.
Is AI in education too broad to be useful as an
organising principle? And is there a better way to be more granular — so that
people working in genuinely similar areas can actually collaborate? Is there
some kind of framework to help educators place their particular work in the
broad field of AI.
To start answering this question, I decided to perform an analysis of the papers submitted to the 2025 SEFI annual conference.
Research questions
1)
To what extent is AI used in
Engineering Education Literature?
2) What is the breadth of use of AI in Engineering Education Literature?
Methodology
A two-stage analysis was performed on the papers presented at the 2025 European Society for Engineering Education (SEFI) Annual conference.
Stage 1: Filter
All papers from the conference were obtained using the conference
website. A text search was performed on each paper to determine if
any of the following target words were present.
●
AI
●
Artificial
●
Intelligence
●
LLM
●
Generative
●
GenAI
●
ChatGPT
●
Copilot
● Machine learning
If a paper contained one or more of these target words, it was retained for stage 2.
There were a number of instances where target words were present in the papers, but were not considered to be part of the subject matter of the work. These included instances such as where one conference theme was... “Digital Education and AI” and some papers cited this theme without including any AI (i.e. they were only digital education), the target words appeared only in the references or were included only in a list of course/topic or module titles.
Stage 2: Categorize
Each of the papers retained for Stage 2 was read and a very short summary of the AI content was written. A thematic analysis was performed on the summaries to attempt to identify the categories of AI in Education. As the number of papers was not large, this was done manually.
Results
The target word search was performed on 174 papers. Of those, 39 (22%) contained one or more of the target words in the body of the document. Of the 39 papers retained for Stage 2, 16 (41%) were considered to not be papers about AI in education, but rather papers placing their work in an educational context that includes AI. For example, a paper about a digital tool to assess thermodynamics learning proposed future work:
"opens up new possibilities for learning analytics using statistical methods, parametric modelling, and perhaps even machine learning in future."
The remaining 23 papers (59% of the Stage 2 papers, 13% of all 174 papers) were considered to be papers involving AI in Engineering Education. From these, the following themes emerged.
|
Theme |
Number |
Interesting
Notes |
|
Understanding
usage and/or perceptions of AI |
8 |
One paper
was about use of AI by staff rather than students |
|
AI as a
tool to be more efficient |
5 |
One paper
has a crossover with the assessment theme |
|
Auditing
current/future skills required in AI |
3 |
One paper
reviews requirements for the public sector |
|
Ethics of
AI |
2 |
Both
papers used film/sci-fi scenarios |
|
Teaching
concepts of AI |
2 |
One paper
was about a course in designing AI hardware |
|
Assessments |
2 |
One paper
was designing assessments robust to AI, the other embracing AI. |
|
Using AI
as a study companion |
1 |
A tuned
chatbot was provided by staff to help students reflect. |
Thematic areas that were not identified in the analysis are:
●
Education of how to use AI
●
Updating curriculum (what is no
longer relevant)
●
How the use of AI is changing
behaviour/learning
This is surprising, given the anticipated disruption of AI to the education sector.
As an aside, the target word search also identified author declarations about the use of AI in the preparation of manuscripts or processing of data. This was not a mandatory requirement for the conference. 21 papers (12%) declared using AI as part of the research work, and 1 declared not using AI.
Conclusions
Despite the widely considered belief that AI will have a significant impact on Engineering Education, this is not significantly represented in the extent to which the subject is presented in the papers presented at the SEFI 2025 annual conference. This could indicate that the conversation and hype is yet to trickle down into practice.
It is also apparent that the emerging themes of AI use focus on a breadth of applications. Returning to where I started: special interest groups are only valuable if their members share meaningful common ground. The analysis above suggests that "AI in Education" is not a theme - it is a container. A researcher studying student perceptions of AI has little to discuss with someone designing AI-robust assessments, who in turn has little overlap with someone auditing the skills that AI is making redundant. If seven distinct themes can emerge from just 23 papers at a single conference, it is likely this phenomenon is more widespread within the sector. Perhaps the honest conclusion is that the special interest groups need to resist the gravitational pull of AI as an organising principle, and instead let the more granular themes do the work. Otherwise, we risk building communities where everyone is talking about AI, but nobody is actually talking to each other.
When citing this work, please use the following
citation:
Garrard, A. (2026). “Everyone’s talking about AI: The breadth of AI in
Engineering Education”. Centre
for Engineering Education Blog, The University of Sheffield,
Sheffield, UK. March 2026. https://www.ceesheffield.co.uk/2026/04/everyones-talking-about-ai-breadth-of.html