An Artificial Intelligence Competency Framework for Teachers and Students: Co-created With Teachers

The integration of artificial intelligence (AI), particularly generative AI (GenAI), into education systems is reshaping the paradigms of teaching, learning, and assessment. In teaching, AI can assist educators by automating administrative tasks (Zawacki-Richter et al., 2019), providing data-driven insights into student performance (Xie et al., 2019), enabling more focused and effective interventions, and generating lesson plans. For students, AI-driven tools and GenAI platforms offer personalized learning experiences, adapting content and pacing to individual learner needs (Baker & Smith, 2019), supporting knowledge and skill development, and assisting in bridging educational gaps across socioeconomic groups (Chaudhry et al., 2022). AI is revolutionizing assessment by offering realtime feedback (Heffernan & Heffernan, 2014), reducing biases (Zhou & Brown, 2020), and facilitating adaptive testing (Van der Linden & Glas, 2010). However, the use of AI in education also raises concerns about ethics, data privacy, and algorithmic bias (Akter et al., 2021; Patel, 2024). According to Chaudhry et al. (2022), students must be equipped with the awareness and understanding necessary to navigate a world increasingly influenced by AI, focusing on developing competencies for real-world educational scenarios. These scenarios should be co-designed with different stakeholders of AI in education, including teachers, ed-tech experts, and AI practitioners. UNESCO (2019) has emphasized the need to ‘dynamically examine and define the roles and skills required of teachers’ to work in education environments saturated with AI and recommends that organizations ‘take institutional actions to improve AI literacy across all layers of society’ (UNESCO, 2019). Therefore, it is imperative to determine the requisite AI competency for teachers and students. Central to this inquiry are the knowledge, skills, values, and attitudes that underpin proficiency in navigating an AI-saturated environment. To ensure that this framework is relevant for teachers in their work, they need to be included as full partners in its development.

Hence, this study aims to (1) identify the necessary competencies for teachers and students in an AI-saturated world, including its essential knowledge, skills, values, and attitudes and (2) co-create and validate an AI competency framework for teachers and students with teachers as research partners.

The widespread use of AI in recent years offers transformative changes in education by providing innovative tools and approaches that can enhance teaching, learning, and assessment (Chakravorti, 2022; Gill et al., 2024; Grubaugh & Levitt, 2023). AI is broadly defined as computer systems capable of performing complex cognitive tasks that are typically considered to require ‘human’ intelligence. These include perception, reasoning, problem-solving, natural language communication, and learning (Russell & Norvig, 2020). Traditional AI systems were predominantly focused on processing large and complex data sets and using restricted rule-based mechanisms to generate responses. GenAI marks a radical shift in its ability to create diverse, complex, and bespoke content, from textual chat responses to software code and movies (Cahane & Shwartz-Altshuler, 2023). This leap is fueled by breakthrough machine learning algorithms trained on unprecedentedly large and diverse data sets. The impact of these new technologies is accentuated by the availability of highly accessible interfaces, making the most powerful technology available to laypersons and even children. The shift from descriptive to creative capabilities has led to the production of original texts, media, and software code, blurring the line between human and machine creativity. This natural interaction with GenAI models promises to deepen abstract thinking and foster the development of complex ideas, revolutionizing human-computer interaction, expanding the demographic of users, and altering public perception of AI’s role in society (Cahane & Shwartz-Altshuler, 2023). A comprehensive AI competency framework should include reference to all aspects of the use of AI.

The transformative impact of AI requires the definition of an AI competency framework for teachers and students. Long and Magerko (2020) defined AI competency as a set of knowledge, skills, values, and attitudes that enable people to critically evaluate AI technologies, manage communication, collaborate effectively with AI, and use it actively in an AI-saturated world. Their framework articulates five core aspects: understanding AI’s nature, gauging its capabilities, dissecting its operational mechanisms, AI ethics, and recognizing human interaction with AI. The framework promotes discernment of AI, including recognition, intelligence understanding, interdisciplinary considerations, and distinguishing between general and narrow applications of AI. Furthermore, it challenges learners to evaluate AI’s potential and limitations. Operational mechanisms are highlighted through competencies concerning AI’s representational systems, decision-making processes, and critical data interpretation and are underscored by the necessity for explainability. The framework contemplates the human dimension of AI, emphasizing the need for programmability, ethical considerations, and the impact of cultural values on the perception and utilization of AI. Intrinsically, the framework insists on ethics as a fundamental competency, thereby interweaving moral contemplation into every facet of AI literacy. This structured yet flexible approach aims to cultivate an informed, critical, and ethically conscious engagement with AI technologies within educational paradigms.

UNESCO has identified ‘the emergence of a set of AI literacy skills required for effective human-machine collaboration, without losing sight of the need for foundational skills such as literacy and numeracy’ (UNESCO, 2023). In September 2023, UNESCO introduced two evolving AI competency frameworks, grounded in human rights and aimed at protecting human dignity, privacy, and enhancing human activities. These frameworks offer teachers a flexible structure to adapt to various classroom environments. The teacher’s framework categorizes 18 essential qualifications for teachers across three levels (acquisition, deepening, and creation) within five thematic areas. For students, the framework outlines essential competencies for engaging with AI in education, emphasizing ethical use, AI methodologies, and design principles. These frameworks were developed through incorporating Member States’ insights on AI curriculum implementation, international consultation meetings, and a comprehensive peer review process involving both field experts and UNESCO colleagues; however, they were not developed in collaboration with teachers. Consequently, although these frameworks identify key areas for development in AI competencies for teachers and students, they fall short in detailing specific skill sets, highlighting a gap in addressing the practical implications of AI in education.

Within the landscape of AI Competency Frameworks for Teachers, the UNESCO Framework represents one possible approach. A critical review by Mikeladze et al. (2024) identifies five distinct types of AI competency frameworks—(1) Existing Competence Model-Oriented Frameworks: These frameworks integrate AI competencies into established theoretical models or general reference frameworks, such as the Technological Pedagogical Content Knowledge (TPACK) and UNESCO’s ICT Competency Framework for Teachers (ICT CFT). This approach leverages validated structures to embed AI competencies into well-known domains (Mishra & Koehler, 2006; UNESCO, 2018); (2) Competence Areas/Domain-Oriented Frameworks: This category includes frameworks that introduce new thematic areas or knowledge domains specific to AI, providing flexibility and adaptability to incorporate emerging areas in AI education. These frameworks are designed to integrate specific new skills or attitudes necessary for educators to engage effectively with AI technologies (Olari & Romeike, 2021); (3) Process-Driven Frameworks: These frameworks emphasize the dynamics of AI applications in the classroom, focusing on how educational processes can be enhanced or transformed through AI integration. They explore how these processes can be translated into competencies or how AI competencies can address these educational dynamics, often using conceptual models to define the interactions between human educators and AI systems (Holstein et al., 2020); (4) AI Systems-Driven Frameworks: Tailored around specific AI systems or models, these frameworks categorize skills based on operational, user, and utilization competencies necessary for effectively managing and interacting with AI systems within educational settings. This approach is particularly useful for educators who need to understand and operate specific AI tools (Carolus et al., 2023); and (5) Competence Level-Driven Frameworks: Offering a multidimensional approach, these frameworks weave various levels of AI competencies into a comprehensive structure. They address a broad spectrum of AI-related skills, from foundational understanding to advanced application and innovation, often involving tiers or levels of competency development (Sattelmaier & Pawlowski, 2023).

By and large, all of these are variants of idealistic, normative top-down approaches: they stem from expert conceptualizations and not from practitioners’ concerns. To our knowledge, none of the suggested frameworks had been developed or validated with teachers as partners. The theoretical background emphasizes that the foundation for an AI competency framework integrates skills, knowledge, attitudes, and values to navigate future challenges in a world saturated with AI, enhancing personal and societal well-being. Due to the high impact of AI and its transformative power on the educational system (Mikeladze et al., 2024), and as teachers are the main change agents integrating technology and pedagogy (Hurtado-Mazeyra et al., 2022), we assert that any framework aiming for broad adoption must engage teachers as full partners in its design. This research aims to co-create and validate an AI competency framework for teachers and students with teachers as research partners. The main research question is: what is the AI competency framework needed for teachers and students in an AI-saturated era, and what are the skills, knowledge, values, and attitudes that are needed?

This research was conducted by the Institute for Applied Research of AI & Education at the R&D Division, Innovation and Technology Administration of Israel’s Ministry of Education (MoE) in collaboration with the Pedagogical Secretariat and Administration of Teaching Staff following MoE policy. This research is part of an extended study aiming to identify an AI competency framework for teachers and students and the ways to implement it in the Israeli education system as a whole. In this article, we focus mainly on the consolidated AI competency framework and discuss preliminary findings regarding the implementation practices.

The study cohort included 43 teachers and 14 school principals from 14 junior high schools (grades 7–9). At the outset of the 2023–2024 academic year, an open call was circulated widely by the MoE, inviting practitioners proficient in AI to join as research partners. A steering committee, including representatives of key directorates in the MoE was formed. This committee assessed candidates based on predefined criteria published in the MoE call and website. The selection criteria aimed for a balance of the social-geographic location, religious affiliations (Jewish, Arab, Druze, and other minorities), and religious observance levels, drawing on Gonen-Avital (2016) and Rebhun (2023) to justify the sample’s composition. Such diversity was not an end in itself but was instrumental in testing the validity and cultural adaptability of the derived competence model. The inclusion of junior high educators reflected UNESCO’s guidelines on AI tools in education, which recommend an age threshold of 13 years (Miao & Holmes, 2023).

The teachers were diverse in their teaching subjects, with seven teachers teaching science and five teachers teaching English as a second language. Three teachers each taught Mathematics and Computer Science. Additionally, several teachers held additional positions, such as pedagogical coordinator, and specialized roles in robotics and environmental sciences.

The study was conducted during the 2023–2024 academic year. The interactions with teachers occurred virtually, using Zoom for six plenary meetings and several individual and small group meetings. Throughout the study period, the participants employed AI tools within their educational settings, evaluated and optimized the framework, chronicled their observations in research diaries, engaged in personal interactions with the research team, and maintained continuous dialogue with their peers through a dedicated WhatsApp group.

This research aims to create a theoretical, conceptualized, and practical AI competency framework that will be adopted by teachers in Israel. We employed a design-centric research-practice partnership (DC-RPP) methodology, combining design-based research (DBR) and RPP methods.

DBR is a systematic but flexible methodology intended to tackle complex, real-world problems and improve educational practices through iterative analysis, design, development, and implementation, based on collaboration among researchers and practitioners in real-world settings (Anderson & Shattuck, 2012; Collins et al., 2004; Mor & Winters, 2007). In each iteration, educational practices and theories are developed and refined. It is a change-oriented methodology, with a strong emphasis on how effective learning and educational innovation can be driven. DBR involves a dual commitment to producing both theoretical and practical products, with researchers deeply involved in the research context, thus eschewing traditional notions of uninvolved researchers and challenging the relevance of dependent and independent variables in such a dynamic setting (Collins et al., 2004; Mor & Winters, 2007).

As teachers are in charge of implementing the framework in their classrooms, their partnership is crucial. To reach these goals, we adopted an RPP paradigm. RPP is a collaborative, mutually beneficial alliance between educational researchers and practitioners (Kali et al., 2018). Coburn and Penuel (2016) emphasized that RPPs focus on mutual engagements, where the work is jointly negotiated, and intentional strategies foster these partnerships through designed roles, routines, and protocols. This paradigm benefits researchers by grounding their work in practical educational challenges and enhances practitioners’ implementation strategies in the classroom by providing evidence-based interventions.

In this study, the emphasis on the teachers’ participation in the research process and its iterative nature led to the adoption of the DC-RPP method. DC-RPPs represent a collaborative endeavour that bridges the gap between theoretical research and practical application, with a strong emphasis on innovative design methodologies (Kali et al., 2018). These partnerships are characterized by their focus on addressing real-world problems through the lens of design, facilitating a symbiotic relationship where design principles and practices inform research and vice versa. Such collaborations typically involve multidisciplinary teams, including researchers and practitioners, working together to co-create solutions that are both theoretically sound and practically viable. The design-centric approach within RPPs underscores the importance of user-centred design, iterative prototyping, and the use of design thinking as a problem-solving framework (Brown, 2008; Norman, 2013). This process ensured that the AI competency framework was both grounded in theoretical underpinnings and responsive to the practical needs and experiences of teachers in primary education settings.

The study adopted a mixed-methods approach, with an emphasis on ethnographic and auto-ethnographic tools, calibrated with automated and semi-automated analysis of surveys, interviews, observations, and content analysis.

Throughout the study, teachers meticulously chronicled their engagement with AI tools and student interactions via research diaries (Amiel & Reeves, 2008). Furthermore, group meetings and individual meetings between researchers and the teachers were conducted, and meeting notes were documented and analysed.

The experiment followed a structured, multi-phase timeline. At the beginning of the school year, in September 2023, a public call for participation was issued, resulting in the selection of 14 diverse junior high schools, including secular, religious, Arab, Druze, and Haredi institutions, along with their school principals and 43 teachers.

This research amalgamated top-down insights and knowledge derived from the scholarly literature, alongside established frameworks from renowned organizations such as OECD, UNESCO, and the INGC. Concurrently, it incorporated bottom-up knowledge generated collaboratively with junior high teachers and school principals. This participatory aspect was crucial in shaping the framework and ensuring its applicability and ecological validation within classroom settings.

The experiment officially began in November 2023 with an introductory plenary session, where the objectives and scope of the experiment were outlined and participants were exposed to different AI tools. Over the next six months, from November 2023 to April 2024, six plenary sessions covered various critical topics such as the basics of AI, ethical considerations, AI in education, and practical implementation strategies. These sessions were complemented by individual school meetings where tailored support and guidance were provided to integrate AI in classrooms effectively. Through all research stages, teachers documented their AI-related classroom activities and experiences, which were analysed to develop the first version of the AI competency model. This initial model was further refined after each plenary session based on feedback and additional data analysis. The final phase included concluding sessions where the outcomes and findings were shared, and a comprehensive analysis of the collected data was conducted.

Interviews conducted with school principals and teachers, ahead of the experiment, indicated that the research partners’ familiarity with GenAI tools was varied, mostly beginner level. Therefore, the first iteration, began in November 2023, focused on establishing basic skills that will enable research-partners working with GenAI, prompting and generating images. After establishing the basic knowledge about the use ofAI and its tools, the research team encouraged the teachers to use GenAI tools in their classroom while documenting their experiences in the research diaries. The research team analysed the data and managed to create the first version of the AI skills framework, which included four skills and different domains for each skill.

The second iteration focused on AI skills. During these meetings, the teachers presented the AI-driven activities they used in their classroom. Afterward, they were introduced to the first version of the AI skills framework that emerged from their research diaries. Throughout the meeting, teachers have used AI tools for three purposes: (1) to map AI skills that were embodied in their first classroom activities, (2) to examine the skills that had not yet been implemented in their classroom following the presented skills framework, and (3) to design a complementary activity to teach AI skills they had not yet implemented. After the second iteration, teachers were asked to implement the AI skills activity they designed and to comment on the first version of the framework. Personal meetings were conducted between teachers and researchers. The analysis included reviewing teachers’ comments and suggestions to develop the second version of the framework to encompass and outline the knowledge, attitudes, and values that emerged from the data. After a careful analysis of their second report, focusing on intentionally teaching AI skills, the issue of ethical concerns stood out as teachers were dealing with it when implementing AI in their classrooms.

The third iteration focused on AI ethics and the ways to approach this issue when implementing AI tools in the classroom. After teachers presented their progress, thoughts, and concerns, AI ethical principles and issues were presented, with a focus on the ethical skill as it occurs in the second version of the AI skills framework. The teachers were grouped with other teachers participating in the research and were asked to plan activities aimed at enhancing ethical use in navigating the AI landscape and taking into consideration the requisite skills, knowledge, values, and attitudes essential for students. The teachers were provided with different materials regarding AI ethics, including Israel’s MoE guidelines. The teachers were then asked to practice this activity in their classroom and document it in their diaries.

Between the third and fourth iterations, an analysis of the research diaries revealed that initially, teachers concentrated on introducing the broad principles of ethics before specifically addressing the nuances of ethics in AI. Furthermore, the diaries delineated various components constructing the core abilities within ethical skills, identified as the third layer in the skill development hierarchy, leading to the third version of the framework. Subsequent individual meetings between teachers and researchers illuminated the necessity for a step-by-step structure for teachers to fully comprehend and integrate ethical skills. These insights were essential for the skills framework structure. At this point, the research team and the teachers acknowledged that each skill consisted of different core abilities, and each ability had several components. After receiving teachers’ feedback, the framework was distilled and consolidated into four skills and different domains: knowledge, values, and attitudes. The teachers also raise the need for a maturity model to improve their work with the students.

Adapting Bloom’s taxonomy (Bloom, 1956), which suggests a hierarchical model that classifies different levels of cognitive skills and learning objectives, four levels of assessment were proposed: remember, understand, apply, and create. These levels paved the way to a maturity model and suggested actions that foster the skill. The fourth iteration concentrated on developing a maturity rubric, and the fifth extended the implementation of AI to encompass all school staff. Due to space limitations, the development of maturity rubric and school implementation will be discussed briefly.

While existing frameworks share some common elements with our approach, there are significant differences in both development methodology and resulting structure. UNESCO’s framework (2023), for example, was developed through expert consultations and policy analysis, without direct classroom implementation or teacher co-creation. Similarly, the frameworks developed by Olari and Romeike (2021) and Holstein et al. (2020) emerged primarily from theoretical analysis and expert input rather than practitioner collaboration. This top-down development approach, while valuable, may not fully address the practical needs of classroom teachers.

Our framework’s categories and sub-categories, shaped by direct teacher involvement, reflect a more practice-oriented approach. First, while other frameworks typically separate teacher and student competencies, our unified approach acknowledges the concurrent learning journey of both groups in the AI era. This theme is elaborated below. Second, our framework uniquely emphasizes AI agency: proactive and valuegenerating utilization of AI as a distinct skill category, moving beyond the typical focus on technical operation and understanding found in existing frameworks. The sub-categories within each skill domain were directly informed by classroom implementation, resulting in more button-up practically oriented components.

The impact of teacher collaboration on our framework’s structure is particularly evident in two key areas. First, the inclusion of ‘Metacognition and Reflection in AI Use’ as core components within the AI agency skill domain emerged directly from teachers’ experiences, as exemplified by one teacher’s observation: ‘The most important thing is critical thinking and not taking for granted what the bot wrote, even if the answer seems logical to the students’ (Teacher H). Second, the ‘Ethical use of AI’ category and its sub-components were directly shaped by teachers’ experiences with ethical dilemmas in classroom settings, rather than being solely derived from theoretical principles. Most frameworks refer to the ethical dimension, but their primary focus is on technical competencies or broad ethical guidelines. Our framework explicitly recognizes these practical dimensions of AI integration, reflecting teachers’ real-world experiences and needs.

The role of AI competency within the Israeli National Pedagogical Policy (Israeli Ministry of Education, 2021a; 2021b) and INGC (Israeli Ministry of Education, 2021b) is fundamental. The purpose of INGC framework is to shape the character of the Israeli graduate and ensure their readiness for the ever-changing world. The INGC outlines 13 distinct literacies and skills. However, given AI’s unique characteristics and pervasive impact, it is recommended to shift the approach: from perceiving technology as a tool to recognizing the existence of an innovative relationship between users and socially significant technology (Mishra & Oster, 2023; Shamir & Levin, 2022). This technology development requires reevaluating the evolving learning framework to include AI competence as the 14th competency, aiding in the formation of a clear, regulated, and integrated pedagogical policy.

As opposed to the different frameworks that differentiate between teachers and students, we argue that a unified AI competency framework for both teachers and students is more appropriate. In this initial stage of AI and GenAI penetration into education systems, we are all teachers and students simultaneously. Despite its seemingly dizzying pace, it is being studied concurrently by both teachers and students. Therefore, this framework is initially designed for learners of all ages and roles. As was observed by the teachers who participated in the research, the same fundamentals of AI literacy and ethics should be common to all users. During the experimental phase process, the conventional hierarchical teacher-student role distinctions become vague as both are beginning at the same place, building their knowledge and skills together. This scenario has the potential to foster the sharing of knowledge and understanding (Mercer et al., 2020) and encourage collaborative thinking in teaching and learning dynamics. However, teachers and students can use the framework differently. For teachers, the framework emphasizes advanced modules on curriculum integration, pedagogical strategies, and ethical decision-making in AI applications, reflecting their role in managing classroom dynamics and adapting teaching methods (Su et al., 2022). Conversely, for students, the focus is on understanding AI principles, ethical usage, and practical interactions with AI tools, fostering critical thinking and responsible use (Miao & Holmes, 2023). This dual approach ensures that while the framework serves a common purpose of promoting AI competence, it respects and addresses the distinct educational and developmental needs of each group (Kali et al., 2018).

The model presented in this manuscript is dynamic and will evolve alongside advances in AI tools and their applications. This perspective aligns with the assertions of scholars such as Russell and Norvig (2020), who emphasize the rapid evolution of AI technologies and the consequent shifts in required competencies for effective utilization. The advancement of AI necessitates a continuous reassessment of the competency required to navigate the AI landscape and cultivate a flexible and forward-looking model (Bostrom, 2014). As AI tools continue to evolve, the necessity for updating policy and regulation becomes crucial. This approach ensures that the model remains relevant and effective in cultivating the requisite competency for engaging with AI technologies, both in their current state and as they evolve in the future.

As teachers implemented the outlined model in their classrooms, the need for a self-evaluation skill model for teachers and students became clear. Throughout the study period, teachers were invited to participate in creating an assessment rubric to evaluate their students’ AI proficiency level. The rubric defined the skills expected from junior high students at various levels, drawing from Bloom’s taxonomy (Bloom, 1956). Due to space limitations, a detailed elaboration is not provided in the manuscript.

The development of a framework for AI competency for teachers and students represents a significant step toward integrating AI into educational environments. However, the value of this framework is contingent upon its active utilization by teachers and students in their daily educational activities, including learning, teaching, and assessment (Zheng et al., 2020). To ensure the effective integration of the AI competency framework, it is imperative to devise a school assimilation model. This model would serve as an organizational tool, helping schools adopt and comprehensively validate the framework. The assimilation model should be designed to support schools in navigating the complexities of implementing AI-driven educational strategies, thereby fostering an environment conducive to the practical application of AI competencies in real-world educational scenarios (Wang, 2019). The formation of these assimilation models is still in progress; as such, they are not included in this manuscript. However, they represent a significant milestone toward implementing the framework for AI competency. The models are crafted with adaptability in mind, enabling them to be tailored to align with each educational institution’s distinct cultural attributes. This customization is vital, as it recognizes the diversity of school cultures and emphasizes the necessity of a context-specific approach to implementing AI competencies in educational settings (Hall & Hord, 2015; Zhao & Frank, 2003).

The ongoing study presented here aims to define an AI competency framework for teachers and students. AI competency is a novel evolving field, as GenAI itself is rapidly changing. The changes that could take place in the coming years must be examined. Moreover, competency consists of a combination of skills, knowledge, values, and attitudes. At this stage of the research, we were able to detail a model of the core abilities and dimensions of the four skills that were identified; however, the definitions of values, attitudes, and knowledge are still in the process of creation. Future work should present a full model that includes all four components and the connections between them. At the next stage of the research, sharing the findings with other stakeholders (e.g., government agencies, and local, national, and international education authorities) and collaborating with them will be essential. Additional work needs to be done to elaborate the model specifically for particular age groups and communities with unique characteristics, and to fine tune the corresponding assessment rubrics.