Research Article
Creating and Sustaining the Judgement Space: Professional Judgement and Dialogic Inquiry in AI-Supported Higher Education
Synthesis: Ji and Mercer (2026) argue that faculty development in AI-supported higher education has been organized around the wrong question. The international frameworks they survey — UNESCO's AI competency framework for teachers, the OECD teaching compass and its 2026 summit report, DigCompEdu, and the BEME review of faculty development — converge on what individual faculty should know and be able to do, and say almost nothing about the conditions under which professional judgment is actually exercised. The paper names that condition the judgment space: the pedagogical relationship in which judgment is not only used but articulated, examined and sustained. In AI-supported higher education the space becomes three-party, because AI-generated evaluations enter the interaction alongside teacher and student. Working from Biesta's account of educational purpose across qualification, socialization and subjectification, the authors argue that AI restructures the evidence base unevenly, capturing qualification evidence far more readily than evidence of socialization or subjectification, and that this asymmetry privileges responsive practice unless pedagogy is designed to enrich the space. Their framework is the dialogic inquiry triangle — inquiry INTO, THROUGH and FOR dialogue — extended to the three-party interaction, from which they derive six structural principles and a professional judgment audit for faculty development.
No primary data: the paper is conceptual and reports no participants, instruments or dataset. The authors state that the judgment space, its properties and the six principles are theoretically derived and offered for empirical investigation, and the empirical studies they cite were not designed to test them.
Key Findings
- The frameworks specify competencies and leave the conditions untheorized. Ji and Mercer read UNESCO's AI competency framework, the OECD teaching compass, the 2026 International Summit on the Teaching Profession report, DigCompEdu and the BEME review as variations on one move: a list of what individual faculty should know or be able to do. The OECD report asserts that AI should not replace human judgment, relationships or responsibility, but does not say what pedagogical conditions would make that assertion true. That gap, not another missing competency, is what the paper treats as the problem for faculty capability.
- Professional judgment is foundational, not one competency among many. Following Biesta, the authors define it as the situated, normative practice of weighing what is educationally desirable rather than what is technically effective. They add Allal's dual character — judgment is at once a cognitive act of interpreting evidence and a socially situated practice shaped by institutional, collegial and disciplinary context — and Biesta's two constraining forces: the culture of measurement, which narrows judgment to what can be quantified, and the "what works" agenda, which converts normative questions into technical ones. Higher education adds the student-as-consumer model and, now, AI-generated evaluations to that pressure.
- The judgment space has three properties, and AI makes it three-party. It is not a room, a timetable slot or a degree of autonomy, but the quality of interaction through which judgment is articulated and examined. It is relational (it exists between participants), normative (it concerns purposes across all three of Biesta's domains, not only qualification), and conditional (institutional, technological and pedagogical structures either enrich or impoverish it). Its relation to professional judgment is recursive: judgment is exercised in the space and reconstitutes it for the next exchange. Because students use AI independently and outside the pedagogical relationship, their cycles of use re-enter the space from outside.
- AI restructures the evidence base asymmetrically. Qualification-domain evidence — correct answers, error patterns, knowledge gaps, performance metrics — is exactly what AI captures best, because quality criteria there can be more completely specified in advance. The meaning of a student's hesitation, risk-taking or shift in perspective cannot be pre-specified that way, and requires a teacher who knows this student, this discipline and these purposes. The authors read Fan et al.'s (2025) randomized experiment — essay scores up, no corresponding gains in knowledge acquisition or transfer, a pattern the original authors call "metacognitive laziness" — as evidence consistent with the asymmetry: qualification performance improved while independent self-monitoring did not. Alghamdi and Alghizzi's review of 37 empirical studies points the same way, finding AI's gains in responsiveness and consistency persistently paired with concern about weakened teacher–student relationships. They insist this is not a limitation that better technology will fix.
- The same AI output can bypass judgment or enrich it. The paper's central analytic distinction turns on where professional judgment sits, not on whether AI reached the student without mediation. A student who asks a chatbot for feedback on an essay and acts on it is in a judgment-bypassing configuration. If the teacher has designed a seminar in which students examine that same feedback against the quality criteria the discipline values, the identical output becomes material for deliberation. Judgment operates in design, framing and facilitation rather than at the moment the output appears, which is why the pedagogical framework, not the tool, decides the outcome.
- The distinction that matters is responsive versus responsible. AI is responsive by design: it reacts to data in real time, detects patterns and delivers consistent feedback at scale. Professional judgment supplies what it does not — normative commitment across all three domains, ethical accountability for consequences, and willingness to accept that the right educational decision may be the least efficient one. The authors name the practice that holds responsiveness accountable to responsibility the normative pause: the deliberate decision to sustain a student's productive struggle rather than resolve it, because the struggle serves their development as an autonomous thinker.
- The dialogic inquiry triangle operationalizes the space. Inquiry INTO dialogue builds the shared analytical capacity to ask what an AI evaluation measured, which criteria it applied and what it missed. Inquiry THROUGH dialogue makes judgment relational through exploratory talk and interthinking, where participants give reasons, challenge constructively and reach understanding no individual working alone would reach. Inquiry FOR dialogue orients both toward transformation, moving the teacher from responsive-only to responsive-and-responsible practice and the student from dependence on external evaluation toward judging for themselves. From this the authors derive six structural principles, two per domain, and the professional judgment audit: a three-column instrument asking what the teacher does, what students do and how AI is positioned, with a responsive/responsible contrast under each domain as a diagnostic mirror.
How the study was done
This is a conceptual paper. It reports no sampling frame, no participants, no instrument and no analysis; its data availability statement says plainly that no datasets were generated or analyzed. The argument is built in five moves: establish professional judgment as the core of educational professionalism from Biesta and Allal; formalize the judgment space and its three properties; analyze how AI restructures the evidence available to judgment; present the dialogic inquiry triangle as the pedagogical framework that creates and sustains the space; and derive the structural principles, the audit and a research agenda.
The theoretical materials are Biesta's account of educational purpose and the dialogic inquiry tradition — Alexander, Mercer, Wells, and Littleton and Mercer's account of interthinking. The triangle itself was developed through the Think-Talk Toolbox and has been examined in school-based practitioner research in contrasting settings, which the authors cite as the empirical warrant for the framework's internal relationships. Their engagement with evidence is unusually explicit about status: they separate what is theoretically derived from what is empirically supported, note that the studies by Fan et al. and Alghamdi and Alghizzi provide evidence consistent with the epistemic asymmetry without having been designed to test it, and place the extension to higher education, the three-party dynamic and the audit itself in the untested column. The proposed research agenda follows that order: investigate the space empirically, test which principles matter across disciplines, then validate the audit as a collegial instrument.
What this means for practice
- Faculty developers. Treat AI training as a question about conditions rather than tools. The audit's three columns — what I do, what students do, how AI is positioned — locate where judgment has quietly moved to a tool, and the responsive/responsible contrast under each domain gives a discussion prompt that does not require scoring anyone.
- Instructors. Place judgment at design time. The seminar that has students test AI feedback against disciplinary criteria is the cheapest available way to convert a bypassing interaction into an enriching one, and it works with the tools students already use outside class.
- Program and assessment designers. Build the occasions the framework depends on: sustained productive struggle, and tasks that require students to say why a criterion matters and where their own assessment of the work differs from the AI's. That articulation is the outcome, not a by-product of it.
- Administrators. Procurement and assessment policy decide, in advance, whether AI outputs arrive as definitive evaluation or as one source of evidence among several. The six principles also require action beyond the individual teacher — how tools are bought, how outputs are positioned in policy, and how development equips faculty to exercise the normative pause.
- Researchers. The framework is a set of constructs, not a finding. Comparative case studies across disciplines and design-based work across seminar, laboratory, studio and large-enrollment formats are the tests the authors themselves propose, and the audit should be used as a discussion artifact rather than a measurement until that validation exists.
Limitations
- No empirical test of the framework. The judgment space, its properties, the six principles and the audit are theoretically derived; the authors say so and place them in a research agenda rather than in a findings section. Nothing here has been observed in a higher education setting.
- The evidence base is school-level. The triangle's relationships, including the claim that collaborative deliberation depends on shared analytical capacity, come from practitioner research in schools; whether they hold where students bring greater disciplinary expertise and engage AI independently is untested.
- The empirical citations were not designed for this argument. Fan et al. and Alghamdi and Alghizzi supply evidence consistent with the epistemic asymmetry, not a test of it, and the systematic review evidence on self-regulated learning is mixed.
- Dialogic practice is culturally and linguistically specific. The conditions for exploratory talk may take different forms, or face different constraints, across the traditions in which higher education operates — a limitation the authors raise themselves.
- The framework assumes pedagogical discretion. Where curriculum, assessment and AI procurement are centrally determined, or resources are thin, the scope for designing the conditions described is substantially narrower than the paper's examples imply.
- The audit is unvalidated. It is offered as a reflective tool for collegial discussion, and the question of whether using it changes how faculty design and deliberate about the three-party interaction remains open.
Citation
Ji, Y., & Mercer, N. (2026). Creating and Sustaining the Judgement Space: Professional Judgement and Dialogic Inquiry in AI-Supported Higher Education. Frontiers of Digital Education, 4(1), article 2.