GenAI as a runaway object in higher education: A socio-cultural view on AI-influenced academic practice in mathematics

Synthesis: GenAI as a runaway object in higher education: A socio-cultural view on AI-influenced academic practice in mathematics

Key Findings

A qualitative study of ten mathematics academics at a Swedish university, analyzed through Cultural-Historical Activity Theory (CHAT), examines how GenAI reshapes the interrelated activity systems of teaching and research.

GenAI is conceptualized as a "runaway object" — a technology that evolves unpredictably, disrupts boundaries, and reconfigures roles, tools, and epistemic norms.

Five dominant themes characterize emerging change across both activity systems: boundary fluidity, shifting objects, epistemic concerns, distributed innovation, and reconfigured roles.

In the research activity system, GenAI was most commonly described as a peripheral support tool used to streamline tasks, with most participants reporting it had not yet penetrated the core processes of scientific knowledge development; in teaching, participants described a shift in the teacher's role from content expert toward facilitator and mediator of critical reasoning, with students portrayed as early adopters who bring new tools, answers, and expectations into the classroom.

Adoption unfolded largely through informal and distributed processes, often associated with student use and peer networks rather than institutional policy, and extended across both teaching and research — a pattern the authors read as evidence of the uneven, hard-to-control diffusion characteristic of runaway objects.

Epistemic concerns were especially pronounced in teaching: participants described a tension between the convenience of AI-generated outputs and the deeper purpose of learning mathematics, warning of a "fake sense of being good at something" when polished output replaces genuine struggle, and of surface-level understanding of methods and assumptions.

The findings suggest GenAI is beginning to influence how academic work is carried out and evaluated, contributing to the blurring of established boundaries and to shifts in how academics orient their work.

Study Design & Method

The study draws on qualitative data from ten academics in a mathematics department at a Swedish university. Data were collected in the early weeks of the spring semester of the 2024–2025 academic year (January–March), at an early stage of GenAI adoption; the primary source was semi-structured individual interviews of approximately one hour each, audio-recorded and fully transcribed. Analysis followed a thematic approach inspired by Braun and Clarke (2006), informed by CHAT constructs (Engeström, 1987), combining inductive coding grounded in participants' accounts with deductive, activity-theoretical coding; themes were assessed for robustness through recurrence across the ten interviews. Using CHAT as its analytical lens, the study treats teaching and research as interrelated activity systems and examines how GenAI acts as a runaway object within them. By highlighting early-stage dynamics, the analysis demonstrates the analytical value of this lens for examining how GenAI is being negotiated within mathematics teaching and research in Higher Ed.

Implications for AI in Education

For Higher Ed institutions and Math Education departments, the findings suggest that formal policy is lagging behind practice: because adoption is driven by students and peer networks, institutions may be designing governance for a technology that is already reshaping academic work from below — and participants themselves called for shared institutional frameworks so that "every teacher" is not forced to invent their own approach. The five themes give Faculty Development and institutional strategy a vocabulary for responding — acknowledging epistemic concerns and reconfigured roles rather than treating GenAI as a neutral tool — and for deciding where Educational Policy AI should intervene. The observed shifts in Teacher Role — from content expert to mediator of critical engagement, with students sometimes more fluent in the tools than their instructors — point to concrete faculty-development needs around assessing meaningful engagement and maintaining pedagogical control in AI-supported classrooms.

Limitations

The study is a qualitative, activity-theoretical case study at a single institution within the Swedish higher education context, where relatively high levels of institutional trust and proactive engagement with AI are evident; the observed dynamics therefore reflect a specific configuration of rules, norms, and community relations and are not representative of other settings. Data were collected at one point in time in the early stage of GenAI adoption, so findings are context-bound and in flux — practices that currently appear peripheral, supportive, or informal, particularly in research, may become more embedded or differently regulated over time. The authors also caution that the findings identify empirical patterns consistent with a runaway-object perspective rather than fully demonstrating GenAI as a runaway object in the strong sense proposed by Engeström.

Connected Concepts

Connected Articles

AI TPACK Preservice Math Teachers

Epistemic Proactivity Math

Citation

Bakogianni, D., Liljekvist, Y., & Bui, P. (2026). GenAI as a runaway object in higher education: A socio-cultural view on AI-influenced academic practice in mathematics.