Conference Agenda

Generative artificial intelligence (GenAI) is entering schools at speed. In policy and professional discourse, GenAI is often framed as an innovation that can improve teaching quality, support differentiated instruction, and reduce teachers’ routine workload (Kasneci et al., 2023; Selwyn, 2019; UNESCO, 2023). Yet emerging practice points to a tension that existing research has not fully explained: GenAI does not simply “add capacity” to teaching. Instead, it is taken up inside evaluation-dense school systems where teachers’ work is continuously sorted, compared, and rewarded across multiple tracks - teaching performance, research/lesson-study outputs, administrative tasks, and competitive awards - often under performative and audit-like expectations (Ball, 2003; Power, 1997). In such contexts, GenAI’s effects are likely to be uneven. It offers immediate gains in work that is highly text-based, templateable, and output-oriented (e.g., drafting lesson plans, polishing narratives for competitions, packaging research/teaching outputs), while providing less direct leverage for relational, situated, and responsibility-heavy labour (e.g., classroom responsiveness, long-term pastoral care, ethical judgment in interaction) (Zawacki-Richter et al., 2019). As a result, “efficiency” does not automatically translate into “reduced burden.” It may instead reconfigure what is experienced as worthwhile work by shifting effort toward activities that are more easily optimised and made visible as evaluative evidence, thereby intensifying metric effects and reactivity (Espeland & Sauder, 2007; Williamson, 2017).

This study addresses this puzzle by theorising GenAI as an amplifier of evaluative visibility rather than a neutral instructional tool. It asks how evaluation-driven visibility regimes interact with GenAI’s selective optimisability to generate misalignment pressures: pressures arising when the work that is easiest to enhance and document diverges from the work teachers value as educationally meaningful but harder to render as evidence (Ball, 2003; Espeland & Sauder, 2007). The study centres teacher agency as a situated professional accomplishment rather than an individual trait, analysing agency as interpretive and strategic work under constraint - reallocating time and attention, deciding where and how to use GenAI, maintaining boundaries around “non-delegable” professional judgment and relational work, and managing moral and reputational risk as visibility expectations intensify (Biesta & Tedder, 2007; Priestley et al., 2015).

Three research questions guide the inquiry:
RQ1 What misalignment pressures arise when GenAI selectively increases the optimisability and visibility of particular teacher-work dimensions under multi-track evaluation systems?
RQ2 What distinct agency strategies do teachers develop in response?
RQ3 What professional consequences are associated with these strategies for sustainability (e.g., meaning, stress, perceived control, identity coherence)?

Empirically, China serves as a strategic case because evaluation density and competitive visibility mechanisms are pronounced, making the underlying mechanisms easier to observe (Ball, 2003; Power, 1997). The study does not treat China as exceptional; rather, it uses a “high-intensity” context to surface dynamics that resonate globally as many systems - including in Europe - expand performance indicators, accountability infrastructures, and digital governance tools (Espeland & Sauder, 2007; Williamson, 2017). Methodologically, the analysis combines (1) publicly available policy and institutional texts that specify evaluative evidence and priorities with (2) semi-structured interviews with teachers across school types and career stages. By integrating institutional-level evidence with practice narratives, the study will map visibility regimes, trace GenAI-driven payoff shifts across work domains, develop a typology of agency strategies (e.g., visibility-optimising, boundary-protecting, pedagogically re-centring, and hybrid forms), and link these strategies to differentiated experiences of professional sustainability. The expected contribution is an explanatory account of why similar GenAI tools can produce divergent effects across contexts, offering a transferable framework for European debates on teacher workload, accountability, and responsible AI integration (Selwyn, 2019; UNESCO, 2023).

The growing integration of artificial intelligence (AI) in education worldwide has raised concerns about the potential de-professionalization and de-skilling of teachers, particularly as core pedagogical responsibilities, such as lesson planning, knowledge transmission, and assessment, are increasingly supported or automated by AI-enhanced systems. As these tools begin to generate instructional materials and recommendations at scale, pedagogical control may shift away from teachers, positioning them less as professional decision-makers and more as implementers of algorithmically generated content (Cukurova, 2025). This risk underscores the importance of teacher agency and professional judgment in AI-mediated classrooms.

In this context, a growing body of literature argues that educators require a specific set of AI-related competencies to effectively navigate AI integration in ways that protect their professional role and support pedagogically meaningful use (Ng et al., 2023; Cukurova & Miao, 2024; Cukurova, 2025). Even though there are competency frameworks available for teachers (e.g., Cukurova & Miao, 2024), most existing studies and professional development (PD) initiatives have focused on enhancing teachers’ AI competency without adopting a comprehensive framework (e.g., Vazhayil et al., 2019; Ding et al., 2024). Moreover, many PD programs emphasize the technical aspects of AI and ethical concerns, while giving comparatively less attention to the pedagogical integration of AI into teaching and learning.

To address these gaps, the present study aims to develop preservice teachers’ AI competency by adopting an agency-oriented approach. The AI Competency Framework for Teachers proposed by UNESCO (Cukurova & Miao, 2024) served as the primary guide for the design of the AI competency learning module for preservice teachers. The UNESCO Framework employs a competency-based approach to support teachers in integrating AI into teaching with a human-centered approach and is structured as a two-dimensional matrix comprising five competency aspects (AI foundations, human-centered mindset, AI ethics, AI applications, and AI pedagogy) across three progression levels (Acquire, Deepen, Create). To conceptualize teacher agency, the study adopts Priestley et al.’s (2015) ecological model of teacher agency. The model describes teacher agency as a situated achievement that emerges through the interaction of three dimensions: the iterational dimension (past experiences, beliefs, and knowledge), the practical-evaluative dimension (judgements under present constraints and affordances), and the projective dimension (orientations toward future goals).

Accordingly, the study investigates how preservice teachers’ AI competencies are developed and understood through the ecological model of teacher agency by implementing an agency-promoting AI competency learning module informed by the iterational, projective, and practical-evaluative dimensions of teacher agency. Regarding this purpose, the research questions are:

RQ1: How does participation in the agency-promoting AI competency module (the AICompA Module) improve preservice teachers’ AI competency?

RQ2: How does preservice teachers’ agency develop through the ecological model of agency (iterational, practical-evaluative, projective) in the context of the AICompA Module?

The rapid expansion of generative artificial intelligence is transforming writing practices across educational systems worldwide. In higher education, students increasingly encounter feedback mediated through AI-based tools alongside established peer and teacher feedback practices. These developments have introduced new pedagogical tensions, particularly regarding how feedback is interpreted, evaluated and integrated into revision decisions. While much existing research has examined the technical features or perceived usefulness of AI tools, less attention has been paid to how feedback practices are experienced and negotiated as part of learners’ ongoing educational development.

These challenges are especially salient in second language (L2) writing contexts. L2 writers often rely heavily on feedback to support linguistic accuracy, academic conventions and genre expectations, making feedback a central site of learning. As AI-mediated feedback becomes embedded within L2 writing environments, learners must navigate competing sources of evaluative input, raising questions about authorship, responsibility and the formation of judgement. Such issues are not confined to a single educational system but reflect globally shared concerns in language education.

For pre-service language teachers, engagement with feedback carries additional significance. As learners, they are required to make revision decisions within AI-mediated writing tasks; as future teachers, they are simultaneously developing professional understandings of assessment, evaluation and ethical practice. Learning to use feedback therefore becomes intertwined with learning to judge. Understanding how pre-service teachers interpret and negotiate multiple feedback sources is thus central to contemporary debates on teacher professional learning in digitally mediated contexts.

This study examines these issues through an interpretive inquiry conducted with pre-service English teachers in a Chinese university setting. China represents one of the largest second language education systems globally, where academic writing is shaped by strong assessment norms and expectations of linguistic accuracy. These characteristics render the context analytically valuable for examining how emerging feedback practices are negotiated under conditions of high evaluative pressure. Rather than positioning the site as exceptional, the study treats it as an illustrative case through which globally shared pedagogical tensions can be explored.

Conceptually, the study draws on literature on feedback literacy, evaluative judgement and professional learning. Feedback is understood not as information transmission but as a situated practice through which learners interpret criteria, negotiate meanings and construct responsibility for their learning decisions. From this perspective, AI functions as a mediating artefact within a broader multi-source feedback ecology, interacting with peer and teacher feedback across drafting stages. Professional learning is conceptualised as an interpretive process shaped by engagement with tools, relationships and institutional expectations.From this perspective, learning to judge is not merely an individual cognitive skill but a socially and institutionally situated professional practice.

Guided by this framework, the study addresses the following research questions:

1. How do pre-service English teachers engage with multiple feedback sources, including AI-mediated, peer and teacher feedback, across iterative writing tasks?

2. How do pre-service teachers interpret, compare and exercise judgement in prioritising feedback when making revision decisions in AI-mediated writing contexts?

3. What conditions shape pre-service teachers’ developing professional judgement and sense of responsibility in feedback-mediated learning environments?

By foregrounding professional learning processes rather than technological effectiveness, this study contributes to international discussions on how emerging technologies reshape educational practices. The findings aim to offer insights for teacher education by highlighting the importance of pedagogical designs that support judgement formation, reflective comparison and ethical deliberation in AI-mediated writing contexts.