Conference Agenda

The growing influence of artificial intelligence (AI) on everyday life and professional contexts has increased the awareness of the need for AI-related competences (European Parliament and Council, 2024). This need is reinforced by European recommendations calling for enhanced awareness, training, and educational initiatives in schools (Council of Europe & Commissioner for Human Rights, 2019). For such initiatives to be successful, teachers themselves must possess adequate AI-related competences, which in turn require integration into curricula and the development of suitable measurement methods (Hattie, 2023). Initial initiatives aiming to strengthen AI competences among teachers indicate that the availability of a clear competence model, including all relevant dimensions for teaching and learning with and about AI, is highly beneficial (Seyferth-Zapf et al., 2025).

Many existing models and assessments build on the TPACK framework (Mishra & Koehler, 2006), which focuses on the interplay of technological knowledge and purposeful use, content knowledge, and pedagogical knowledge required for effective teaching. However, the pedagogical dimension, particularly regarding teaching about AI, often remains underrepresented. Current AI-related competence models are either targeted broadly at “non-experts” (Laupichler et al., 2023) and therefore lack pedagogical specificity, or they are tailored to teachers but remain strongly oriented toward technical aspects of using AI, only implicitly addressing the demands of teaching about AI (Celik, 2023).

Additionally, existing models and assessments employ heterogeneous terminology and partially inconsistent definitions of what constitutes competence. To address this, the competence definition by Weinert (2001) is adopted as a theoretically robust foundation. Weinert conceptualises competence as a combination of cognitive abilities and skills, together with motivational, volitional, and social dispositions that enable individuals to successfully manage relevant tasks. This conceptualisation provides a well‑established and internationally recognisable basis for examining AI-related teacher competences. Following this understanding, the present study refers to competence, rather than literacy or competency, in line with Weinert‘s (2001) definition.

Based on this conceptual foundation and the identified gap, the lack of a comprehensive model covering all relevant competence dimensions, the aim of this study is to identify, systematise and integrate these dimensions into a coherent AI‑related competence model for teachers and then operationalise it. Accordingly, the study addresses the following research questions:

- Which existing competence models for AI-related competences exist, and what are their strengths and weaknesses regarding pedagogical, technological, and ethical requirements?

- Which specific competences do teachers need in the context of AI, and how can these AI-related competences be defined?

- How should a comprehensive competence model be designed to capture all relevant requirements related to AI for teachers?

- Which methodological approaches and operationalisation strategies are suitable for establishing such a competence model in a theoretically grounded and empirically validated manner?

The resulting model is intended to provide guidance for teacher education, curriculum development and school improvement. By combining European relevance, theoretical clarity and empirical rigour, the project contributes to a coordinated European dialogue on professional AI competences in education.

Artificial intelligence (AI) technology has been increasingly integrated into K-12 education, offering a novel pathway for optimizing instructional practices and improving students' academic performance (Wang et al., 2023). This transformative trend is particularly prominent in science education, where AI tools unlock revolutionary potential for teaching complex scientific concepts, delivering personalized learning, and conducting interdisciplinary scientific inquiry (Erduran, 2023). This imposes higher demands on science teachers: they are required to not only possess the professional competence to lead AI-driven instructional innovation, but also acutely perceive and effectively address various ethical issues that emerge continuously in AI-enabled teaching practices.

In particular, for early career science teachers (ECSTs) in their first three years of appointment, this shift presents both opportunities and challenges, as their core responsibilities entail adapting to technological disruption while establishing foundational professional competencies (Barrett, 2025). ECSTs must not only rapidly master AI-based teaching tools and methodologies to align with evolving instructional requirements, but also navigate the reshaping of professional roles and competency challenges brought about by technological transformation (H et al., 2025). Their AI readiness serves as a pivotal underpinning for addressing the aforementioned pressures.

Against this backdrop, job satisfaction among ECSTs has grown in significance and merits special attention. Neglecting the interactive relationship between AI readiness and job satisfaction will not only directly dampen ECSTs' initiative to proactively integrate AI into teaching and undermine the implementation effectiveness of AI-enabled education, but also adversely affect their career development trajectories. Crucially, low job satisfaction may lead to early career attrition. Surveys indicate that the natural attrition rate of early career teachers (ECTs) has exceeded 30% in some European countries in recent years (OECD, 2021) — a phenomenon that directly reflects their low job satisfaction and thus jeopardizes the stability and sustainable development of the teaching workforce.

While existing research has examined the association between AI readiness and job satisfaction, it lacks empirical evidence in non-Western contexts. In particular, few studies have systematically explored the mechanism through which four-dimensional AI readiness influences job satisfaction among ECSTs in China. Currently, China's Ministry of Education, Ministry of Science and Technology, and other relevant departments have issued a series of policies to promote and support the full integration of AI into K-12 science education (Ministry of Education et al., 2025). This policy context further underscores the urgency of investigating the practical predicaments faced by this teacher cohort.

Drawing on the scale developed by Karaca et al. (2021) and the Technology Acceptance Model (TAM), this study adopts the localized framework proposed by Wang et al. (2023), which defines AI readiness as teachers' state of preparedness across the cognitive, ability, vision, and ethical dimensions in the context of AI implementation in educational settings. This study aims to investigate ECSTs' AI readiness and its impact on job satisfaction against the backdrop of AI-enabled science education. Clarifying the inherent correlation between the two is essentially the key to addressing the dilemma of the escalation of technological requirements and the imbalance in teachers' professional adaptation, and it holds great significance for safeguarding the professional growth of ECSTs and laying a solid foundation for the development of the science teaching workforce in the AI era.

Background and Rationale

The integration of artificial intelligence (AI) into teacher education has attracted increasing attention from scholars and policymakers worldwide, intensifying calls to define and evaluate teachers’ competencies for an AI-rich educational landscape (Zhai & Nehm, 2023). Within teacher education, a growing body of work has sought to conceptualise teachers’ capabilities to use AI tools effectively, ethically, and pedagogically. Building on the Technological Pedagogical Content Knowledge (TPACK) framework, recent scholarship has proposed AI-TPACK by incorporating AI-related dimensions into teachers’ professional knowledge base (Celik, 2023; Ning et al., 2024). Correspondingly, researchers have developed and validated self-report AI-TPACK scales to capture pre- and in-service teachers’ perceived readiness to integrate AI into teaching (Ning et al., 2024), and these instruments are increasingly adopted in empirical studies as a basis for competency evaluation (Karatas & Atac, 2025; Xu et al., 2025).

However, reliance on self-report instruments raises enduring validity concerns, particularly when the target competency involves complex, design-based practice rather than general dispositions or confidence (Akyuz, 2018). This challenge becomes especially salient with generative AI (GenAI), which can support multimodal science teaching—an area where teachers must coordinate multimodal resources (e.g., text, image, audio, animation) to foster conceptual understanding (Kress, 2001; Nielsen & Yeo, 2022). These conditions make it crucial to examine whether self-reported AI-TPACK aligns with demonstrated competence in authentic, AI-supported multimodal design tasks.

Research context

To examine this measurement gap in an authentic teacher-education setting, this study is situated in a pre-service science teacher education course where participants completed an AI-supported multimodal design task. Specifically, pre-service teachers used generative AI to support the creation of web-based animations and interactive representations intended to teach abstract scientific concepts. The same cohort also completed a widely used AI-TPACK self-report questionnaire, enabling a direct comparison between perceived competence and demonstrated performance within a shared instructional context.

Empirically, the study draws on 116 GenAI-assisted multimodal assignments, which were evaluated using a TPACK-aligned performance rubric across the seven TPACK dimensions, and these scores were matched with participants’ questionnaire responses. To contextualise and interpret patterns of convergence and divergence, we further conducted six focus group interviews with 24 pre-service teachers, exploring how they perceived AI’s role in science teaching and how they evaluated their own capabilities when working with AI-supported design tasks.

Research Questions:

RQ1: To what extent do pre-service science teachers’ AI-TPACK self-reports align with performance-based TPACK scores derived from an AI-supported multimodal design task?

RQ2: What patterns of misalignment emerge across the seven TPACK dimensions?

RQ3: How can the observed misalignment be interpreted in terms of construct validity?

Theoretical Framework

This study integrates TPACK (Mishra & Koehler, 2006) with Kane’s (2013) argument-based validation to examine the meaning and limits of AI-TPACK self-report scores in an AI-supported multimodal design context. TPACK specifies the competence domain of interest—teachers’ integrated use of content, pedagogy, and technology knowledge (CK, PK, TK) and their intersections in designing multimodal science representations. Kane’s framework then examines the use of AI-TPACK questionnaires as a validity claim. First, we examine relations to other measures by comparing self-report scores with rubric-based performance scores from an AI-supported multimodal design task. Second, we use qualitative data as response-process evidence to explain why self-reports misalign with task performance and to refine defensible score interpretations.

International Implications

Globally, AI-TPACK self-report scales are increasingly used to monitor teacher readiness and evaluate AI-focused teacher education. Using data from China—a major, fast-evolving AI-in-education context—this study provides internationally relevant validity evidence. It cautions against relying solely on self-reports when benchmarking teacher competence and supports combining questionnaires with performance-based assessment for cross-country comparisons.

Teachers need to learn more effective teaching methods, and it is important for teachers to have a better understanding of instruction to support their students' learning (Murata et al., 2012). The arrival of artificial intelligence (AI) has increased its integration into education, making it necessary to change the way teachers design, deliver, and think about their teaching practices (Novita, 2025). Despite the increasing importance placed on artificial intelligence in education, it remains unclear how teachers will effectively implement it in their classrooms (Ifenthaler et al., 2024). Teachers cannot manage the rapid evolution of AI on their own; therefore, lesson study is ideal for managing this process more efficiently. Lesson study assists as a powerful professional development model to talk this absence, which requires the inclusion of evolving pedagogical strategies aligned with curriculum standards and student needs, is a critical but time-consuming activity among teachers' various teaching tasks (Novita, 2025). Lewis & Tsuchida (1998) defined lesson study as a teaching method with lesson characteristics that are meticulously planned, monitored, recorded, and discussed by other teachers. Moreover, it is presented through a cycle organized in a collaborative environment, consisting of goal setting, curriculum analysis, lesson planning, supervised instruction, and evaluation and review stages (Murata et al., 2012). Teachers can use inquiry in the process because inquiry enable students to use rational thinking, form statements, and evaluating alternative descriptions (Hofstein & Lunetta, 2003). When examining the literature on science education, the concept of inquiry is defined in a multifaceted way, encompassing processes such as exploring science, making observations, asking questions, examining various sources of information and books, collecting data, analysing it, and interpreting it (Hofstein & Lunetta, 2003). Therefore, this study designed as teachers can integrate artificial intelligence into their lessons, from the learning and application stages to the assessment stage. Stages of collaborative lesson study are as follows (Demir et al., 2012). The first stage is goal setting and planning. Instructional activities are designed to help students achieve the stated goals. Second phase is the instruction phase. At this point, one teacher volunteers to teach the lesson, and the others participate as passive observers. Teachers do not evaluate the quality of the instruction; they focus on the obvious signs of student learning. Teachers review the results of their observations and decide together how the observed lesson can be improved. Following this evaluation session, they discuss their experiences of observing and implementing the lesson and focus on revisions to be made to the lesson. Accordingly, revised lesson is implemented in another class by the same teacher or another member (Demir et al., 2012).The aim of this study was to reveal how middle school science teachers design lesson plans using artificial intelligence-supported teaching tools in the lesson study process, how they integrate these tools into their inquiry-based teaching. The fundamental research question that guided this study was: How do middle school science teachers integrate artificial intelligence tools into inquiry-based teaching designs in the lesson study process? This research is expected to be significant because it examines how middle school science teachers prepare lesson plans using AI-supported instructional tools within the framework of a collaborative lesson study process, and how they adopt an inquiry-based learning approach in this process. The research will comprehensively address teachers' experiences in creating AI-enriched lesson plans, the educational opportunities they encounter, and the challenges they face. The collaborative lesson planning process supports teachers' professional development through collaborative planning, implementation, and evaluation. In this context, the integration of AI tools into the teaching process and the reflection of inquiry-based activities in lessons will be analysed in detail from the teachers' perspectives.