Which information is fair game?
The student data an automated decision should be allowed to draw on — and what it shouldn't touch at all.
Co-design workshop · Online · Call for participation
Deciding what counts as fair in educational AI.
Two students alike in every relevant respect can receive different decisions from the same system. This workshop brings researchers, practitioners, and decision-makers together to work out what fair treatment should require — and to build an instrument that measures it.
Why this workshop
So far, fairness in educational AI has been defined mostly by the people who build it.
A demographic disparity like the one above is one form of algorithmic unfairness, and it can take many shapes across the automated decisions universities make about students. How far that unfairness erodes students' trust is still poorly understood — and deciding what counts as fair has largely fallen to system developers, with little input from educators, institutions, or the students affected.
We take the view that fairness in education warrants broader deliberation than developers alone can provide. This workshop is where that deliberation happens.
What we'll examine
Three questions we keep returning to.
They are far from the whole picture. But across the many decisions we examine, these three come up again and again — and for each case, we work out together what fair treatment should require.
How openly is it explained?
How clearly the decision, and the reasoning behind it, are made visible to the student it affects.
Same for all, or adjusted?
Whether every student is treated identically, or allowances are made for personal circumstances.
What it builds
From the room to a reusable instrument.
What's said in the workshop feeds a survey, the survey is checked by the people who shaped it, and the result becomes a framework others can use.
1
Deliberate the cases
Participants work through the automated decisions that raise the sharpest fairness concerns and weigh what fair treatment should require.
2
Draft the survey instrument
Those ideas become a draft instrument that measures how fair students themselves consider a decision to be — and how that judgement shapes their trust.
3
Return it for review Async · ~30–45 min
The draft goes back to participants in a short follow-up to confirm it reads clearly and realistically before it reaches any student.
4
Administer to students
The refined instrument is put to students, whose judgements of fairness and trust are what the study is ultimately about.
5
Grow into a framework
In time the instrument forms the basis of a broader framework for fairness and trust that can be reused well beyond this study.
Why take part
You help shape the instrument — and you get to use it.
Shape the work directly
Help define a survey instrument and framework for studying fairness and trust in educational AI.
Early access
Receive the instrument and the findings ahead of wider release, to use in your own work.
Acknowledgement & authorship
Contributions are acknowledged in resulting outputs, with co-authorship open to those who contribute substantially.
Connect with the field
Meet researchers, practitioners, and decision-makers across the sector working on the same problem.
Format & who should come
Online, and open to anyone with an informed interest.
- Format
- Online
- Main session
- ~2–2.5 hrs
- Follow-up
- ~30–45 min
- Date & time
- October 15, 20265–7.30pm ACST
We welcome researchers, practitioners, and decision-makers concerned with AI systems in education, as well as people who work closely with the students affected. Formal expertise is not a prerequisite — an informed interest in the problem is enough.
Organisers
Organised by the Human-Centered Responsible AI SIG.
A Special Interest Group (HCR-AI) of the European Association of Technology Enhanced Learning, EATEL. Visit the SIG →
Oscar Deho
Adelaide University
View profile →Olga Viberg
KTH Royal Institute of Technology
View profile →Srecko Joksimovic
Adelaide University
View profile →Background reading
Selected work on fairness perceptions in algorithmic decisions.
A few starting points for the questions this workshop takes up.
1
Marcinkowski, F., Kieslich, K., Starke, C. & Lünich, M. (2020). Implications of AI (un-)fairness in higher education admissions: the effects of perceived AI (un-)fairness on exit, voice and organizational reputation. Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency (FAccT '20), 122–130.
2
Starke, C., Baleis, J., Keller, B. & Marcinkowski, F. (2022). Fairness perceptions of algorithmic decision-making: A systematic review of the empirical literature. Big Data & Society, 9(2), 20539517221115189.
3
Lünich, M., Keller, B. & Marcinkowski, F. (2024). Fairness of academic performance prediction for the distribution of support measures for students: Differences in perceived fairness of distributive justice norms. Technology, Knowledge and Learning, 29(2), 1079–1107.
Help decide what fair means.
Registration takes a moment. We'll send the joining details closer to the date.
Register to take part →