Contact
ContactHU
Menu

What does AI change in universities?

Students and educators have used GoSchool systems on real university courses since 2023. That lets us see not only what people think about AI, but how they learn, teach and work with it.

Real use has already produced research results.

The publications and our current projects start from the same premise: study AI in the setting where people actually use it.

Publications

  1. The Power of Outsourcing

    Árpád TamásiThe European Sociologist, Issue 55

    Drawing on close to forty thousand student messages with the course-integrated AI tutor GoSchool.ai, the essay starts from the premise that outsourcing is a rational strategy, not laziness. Genuine dialogue, where the student's own thinking is put to the test, so far shows up in four percent of interactions — something to build on: it points to the kind of course design that opens room for deeper thinking.

    Open the essay (opens in a new tab)

  2. Exploring the Use of Retrieval-Augmented Generation Models in Higher Education

    Renáta Németh, Annamária Tátrai, Miklós Szabó, Péter Tibor Zaletnyik, Árpád TamásiSocial Sciences & Humanities Open

    The study examined course-specific AI across four courses at two universities. In a randomly sampled set coded by experts, 1.5% of answers were incorrect and 16.5% fell outside the context supplied to the model.

    Open the paper (opens in a new tab)

  3. Using a RAG-enhanced large language model in a virtual teaching assistant role: Experiences from a pilot project in statistics education

    Renáta Németh, Annamária Tátrai, Miklós Szabó, Árpád TamásiHungarian Statistical Review, Volume 7 (2024)

    The statistics-education pilot studied an AI tutor grounded in a textbook knowledge base, connecting educator experience with student focus groups and surveys.

    Open the paper (PDF) (opens in a new tab)

Further completed research

  1. How do students relate to an AI assistant?

    ELTE and University of Pécs · three courses · autumn 2024–2025

    Using anonymised interaction logs, course-context data and student reflections, we studied when students attribute intention, understanding or personality to an AI assistant.

  2. How does AI change learning and educators' work?

    OsloMet · one BA and one MA course · 2024–2025 academic year

    The study examined how students used the assistant for learning and assignments, and how the platform shaped course dynamics, educators' work and expectations about knowledge.

  3. What kind of course design supports reflective learning?

    ELTE Interdisciplinary Social Research Doctoral Programme

    Through design-based research cycles, we studied which choices support student self-regulation and cognitive engagement, and how the educator's role changes alongside them.

Real courses. Use observed over time.

Interactions created while GoSchool is in use show, over time, how students turn to AI and how educators shape the learning environment around it.

What that makes possible
  • analysing natural student interactions
  • following change across weeks or months
  • comparing courses and educator configurations
  • connecting qualitative and quantitative analysis
  • interpreting student and educator perspectives together

Conversations may be used for research and analysis only in anonymised or aggregated form, so that individual users cannot be identified.

We study what changes with AI.

We look at five connected areas, from student use to institution-wide adoption.

  1. Student AI use

    When do students seek explanation or practice, and when do they hand the work off?

  2. Learning and independence

    How do help-seeking, self-regulation and independent problem-solving change?

  3. Educators' work

    Which tasks, course-design choices and forms of control does AI reshape?

  4. Human–AI collaboration

    What role does AI take over time: explainer, practice partner, search tool or substitute?

  5. Institutional adoption

    How can AI fit a university's existing technical, educational and organisational environment?

Research directly shapes what we build.

We do not use research to validate a finished solution after the fact. What we learn from use already shapes the next decisions we make.

Real use teaches us
  • what help people actually need
  • where educators need more control
  • which interfaces and AI capabilities work
  • what works technically but creates no value

Let's investigate a real university question together.

We work with universities, research groups and educators where the real use of AI in higher education can be studied.

Let's talk about research collaboration

Possible collaboration: Joint research · Course pilot · Data collection · Joint publication