Abstract:

As generative AI becomes increasingly embedded in higher education, educators face a growing tension between supporting learners to work productively with AI and resisting tools that bypass learning, erode academic integrity, or impose opaque, surveillance-driven practices. Much current discourse focuses on restricting or detecting AI use, rather than on how pedagogically sound engagement with AI might be intentionally designed.

This practitioner paper presents a reflective account of pedagogy bots within the Studi platform; lecturer-defined, assessment attached AI agents designed to support learner assessments as a co-creative, process-focused activity. Unlike generic AI tools or institutionally imposed systems, pedagogy bots are optional, fully configurable by lecturers, and explicitly grounded in the pedagogical intent of a specific assessment. In this way, they act as 24/7 pedagogical companions that extend educator vision while preserving learner agency. Pedagogy bots are designed not to provide answers or optimise outputs, but to scaffold thinking, reflection, and decision-making as learners work through their assessments over time. By embedding guidance, questioning strategies, and discipline-specific expectations directly into the assessment context, these bots support learners in developing effective study practices while engaging responsibly with AI. This approach foregrounds learning processes rather than final products and reframes academic integrity as something cultivated through transparent, supported engagement rather than enforced through post-hoc detection. Drawing on early use of pedagogy bots in higher education settings, this paper reflects on their impact on both learner experience and educator practice. Preliminary insights suggest that learners experience greater clarity, reduced anxiety around AI use, and increased confidence in navigating complex tasks, while educators benefit from a scalable way to express and sustain pedagogical intent beyond scheduled contact time. Importantly, this design supports inclusivity by offering consistent, always available support that accommodates diverse study rhythms and needs.

Finally, the paper argues that lecturer-defined pedagogy bots represent a future-proof approach to AI-enabled assessment. By anchoring AI use in pedagogical intent rather than specific tools or models, this design remains resilient to rapid technological change while empowering learners to develop transferable judgement, autonomy, and ethical awareness in their engagement with AI capabilities that extend beyond any single assessment or platform.

Keywords:

Pedagogy bots, Learner-centred assessment, Responsible AI, Learner agency, Co-creative pedagogy, Academic integrity, Inclusive EdTech.