University of Amsterdam funds two PhDs

Universities are done treating generative AI as a side issue. The University of Amsterdam just opened two fully funded PhD positions built around a much harder question — what happens to teaching, grading, and fairness when AI tools are everywhere. That matters because campuses have spent the last two years improvising. Policies changed fast, but the evidence base stayed thin. Now UvA is putting real money and four years of research time behind closing that gap. ### What exactly did Amsterdam post? UvA’s Faculty of Social and Behavioural Sciences posted two four-year PhD openings in Educational Data Science. The project sits inside an interdisciplinary research effort on higher education, and the university is looking for candidates who can work across education research, data science, and AI. Applications close on May 7, 2026, and the listed gross monthly salary runs from €3,059 to €3,881 over the PhD pay scale. (werkenbij.uva.nl) ### What will the PhDs actually study? The core topic is not “AI” in the abstract. It is how teaching, learning, and assessment change when universities can observe student behavior through large institutional datasets. The vacancy says the researchers will use online(werkenbij.uva.nl)lts, and evaluations. The stated aim is to study how assessment policies, institutional practices, and GenAI tools shape student success, fairness, and curricular validity. (werkenbij.uva.nl) ### Why is that more interesting than it sounds? Because the hard problem is no longer just cheating. A lot of universities started with that frame in 2023 — can students use ChatGPT on homework, and how do you catch them if they do? But the deeper issue is whether ol(werkenbij.uva.nl)reward AI-assisted fluency over actual understanding, then fairness starts to drift even when nobody is breaking rules. UvA’s project is basically about measuring that drift instead of arguing about it from vibes. (werkenbij.uva.nl) ### Why use university data for this? Because self-reports are weak. Students do not always say when they used AI, how they used it, or whether it changed the way they studied. Institutional data gives researchers a better shot at seeing patterns across courses and ov(werkenbij.uva.nl)is kind of work only becomes useful if it connects behavior to policy. That seems to be the design here. The project is explicitly about teaching, learning, and assessment decisions, not just dashboards. (werkenbij.uva.nl) ### Who are these roles for? The ad is aimed at people with a master’s degree and an interest in education plus quantitative methods. But the bigger signal is institutional. UvA is not hiring one experimental AI fellow to run a pilot. It is funding two full PhDs as pa(werkenbij.uva.nl)ware. (werkenbij.uva.nl) ### Why does the timing matter now? April 2026 is a different moment from the first GenAI shock. Campuses have had time to rewrite rules, test oral exams, redesign assignments, and argue over bans versus disclosure. What they still lack is strong empirical work on whi(werkenbij.uva.nl)ncy policy and more about evidence-backed redesign. (werkenbij.uva.nl) ### So what is the real takeaway? This is a hiring post, not a scientific breakthrough. But it is still a useful signal. A major European university is treating generative AI in assessment as a long-horizon research problem — one tied to fairness, student success, and whether grades still mean what universities say they mean. That is where the conversation is heading. (werkenbij.uva.nl)