Guidelines for learning design and assessment for generative artificial intelligence-integrated education: a unified view

The integration of generative artificial intelligence (GenAI) into learning design and assessment represents a significant evolution in pedagogical practices, necessitating a thorough investigation of international policy frameworks to guide its responsible and effective implementation. Although research on artificial intelligence (AI) in education is not new (Baker et al., 2021; Molenaar, 2022; Wang et al., 2024), with the Artificial Intelligence in Education (AIED) community established for over two decades (Holmes, 2024), recent advancements in GenAI have significantly impacted learning and teaching (Kishore et al., 2023; Jochim and Lenz-Kesekamp, 2024).

GenAI, distinguished by its capability to produce novel content, including text, images and simulations, presents transformative opportunities for educational contexts. By enabling personalised learning experiences, GenAI can dynamically adapt instructional content to individual learner needs, enhance engagement and foster deeper, creative problem-solving. Furthermore, it supports the implementation of adaptive instructional strategies that respond in real time to student progress, potentially optimising learning pathways and outcomes (Nguyen et al., 2024a). Despite these advantages, the rapid uptake of GenAI in educational settings brings complex challenges that must be addressed to ensure ethical and equitable use.

Concerns regarding the fairness of AI-driven decisions, the validity of AI-generated content and the reliability of such content as a basis for assessment highlight critical issues (Giannakos et al., 2024; Fan et al., 2025). Without robust policies, the integration of GenAI risks reinforcing existing educational inequalities (Nguyen et al., 2023). For instance, disparities in access to GenAI tools could widen the digital divide, depriving learners from under-resourced institutions. Additionally, unchecked reliance on GenAI could hinder the development of critical thinking if students are provided with automated answers rather than guided support. Thus, a deliberate approach to GenAI adoption is essential to prevent potential biases, ensure transparency in content generation and establish clear frameworks for evaluating AI’s role in learning assessment. Addressing these issues is crucial for safeguarding educational integrity and fostering an inclusive, equitable learning environment where GenAI serves as a tool for empowerment rather than an obstacle to fair educational practices.

Accordingly, this study embarks on an analytical exploration of policy documents issued by key intergovernmental organisations (IGOs) such as the Organisation for Economic Co-operation and Development (OECD), the United Nations Educational, Scientific and Cultural Organisation (UNESCO) and the European Union. The motivation for this research stems from the growing recognition of GenAI’s transformative capabilities in reshaping learning design and assessment approaches, thereby requiring a detailed understanding of the policy landscapes that govern its application.

In this study, our objective is to systematically map the policy terrain that underpins the deployment and governance of GenAI in educational assessments. By examining related policy documents posited by the international entities, this research aims to examine the core principles, regulatory frameworks and ethical guidelines for learning design and assessments. This endeavour is particularly relevant, given the current scholarly focus on the convergence of GenAI technology and educational practices, a field that is rapidly evolving yet underexplored in terms of policy analysis. In particular, this study aims to address the following research questions (RQs) from the perspective of policy guidelines:

In synthesising the findings from this policy document analysis, the study aims to offer valuable insights for educators, policymakers and researchers. By understanding the policy frameworks that govern the use of GenAI in educational assessments, stakeholders can better navigate the challenges and opportunities presented by this technological advancement. This research, therefore, holds significance not only for the immediate field of educational technology but also for the broader landscape of education policy and practice in the era of GenAI.

GenAI is a subset of AI that generates new content, such as text, images or data, by identifying patterns in existing data sets, thus mimicking or extending human creativity and problem-solving capabilities. Even before GenAI’s proliferation, AI-enhanced systems demonstrated effectiveness in providing personalised feedback to students by analysing quantitative and qualitative data (Fadlelmula and Qadhi, 2024; Wang et al., 2024). Recent studies have highlighted GenAI’s capacity to better deliver personalised, timely and continuous formative feedback, significantly improving student engagement and outcomes (Giannakos et al., 2024; Nguyen et al., 2024a). The integration of AI into feedback systems enables more individualised evaluations of student strengths and improvement areas. However, AI-driven feedback within digital learning environments can present challenges (Giannakos et al., 2024). Studies indicate that AI-generated feedback may lead to overreliance, misinterpretations and negative emotional reactions, highlighting the need for careful algorithm design and continuous refinement (Jin et al., 2024; Fan et al., 2025). Negative emotions such as neglect, frustration, uncertainty and discomfort have been reported among students, emphasising the necessity of designing supportive and constructive feedback systems (Saplacan et al., 2018).

Advancements in GenAI further enhance previous AI feedback systems by providing more human-like interactions. Research suggests human-like features in AI significantly increase learners’ trust, influencing their acceptance and integration of feedback (Lee, 2010; Parenti et al., 2023; Nguyen et al., 2025). Such human-like feedback, incorporating emotional tones and adaptability, addresses limitations of earlier systems. AI applications using natural language and tailored media content significantly boost student engagement and perceived usefulness, enhancing overall learning outcomes (Zhang et al., 2024). Additionally, this approach facilitates empathetic and supportive responses to learner needs. Moreover, recent advancements enable GenAI to offer multimodal feedback, combining text, visual, auditory and other communication forms (Becerra et al., 2024; Nguyen et al., 2024b). Multimodal feedback addresses diverse information-processing styles, making learning more engaging and accessible (Sigrist et al., 2013; Whitehead et al., 2024). By contextualising abstract concepts, multimodal approaches encourage active engagement, fostering deeper understanding and adaptive learning outcomes.

Beyond feedback, GenAI’s advanced content-generation capabilities also support adaptive learning and self-assessment (Waluyo and Kusumastuti, 2024; Xia et al., 2024). AI dynamically tailors educational content to individual learners, accommodating diverse needs and enhancing educational inclusivity (US Department of Education, 2023). This aligns with recent policy guidance from UNESCO, OECD-Education International (2023) and the European Commission, which emphasise equity, fairness and accessibility in AI applications [European Commission: Directorate General for Education, Youth, Sport and Culture, 2022; Miao and Holmes, 2023; Organisation for Economic Co-operation and Development (OECD)-Education International, 2023]. Intelligent tutoring systems (ITS) leveraging AI effectively customise instructional methods and self-assessment tools, enhancing learner autonomy and accountability (Baker et al., 2021; Mousavinasab et al., 2021; Nguyen et al., 2024a).

Formative assessment is another critical application of AI in education, notably exemplified by systems like StuDiAsE (Samarakou et al., 2016). Predictive analytics integrated into formative assessments offer real-time performance insights, guiding personalised learning strategies (Gašević et al., 2023; Wang et al., 2024). This dynamic approach shifts from traditional assessments towards more responsive, student-centred learning environments. Additionally, AI automates student grading, particularly beneficial in online exams, enhancing efficiency and scalability (Khaleel et al., 2020; Messer et al., 2024). AI also aids in test creation within e-learning platforms, demonstrating flexibility in meeting educational demands.

However, the use of AI in student assessment brings about various ethical concerns that warrant careful consideration and mitigation. Some of the limitations and ethical issues include bias and fairness, privacy concerns, transparency and explainability, equity and accessibility, emotional impact on students, security and cheating prevention (Prinsloo and Slade, 2015; Nguyen et al., 2023). Furthermore, realising the full potential of GenAI in educational assessment necessitates focused teacher training and continued research efforts (Gašević et al., 2023; Nguyen, 2025). Educators need to be equipped with the skills to understand and effectively integrate AI tools into their assessment practices. Furthermore, ongoing research is crucial to explore the diverse possibilities and challenges associated with GenAI in educational assessment.

Previous research emphasises the critical need for transparent AI systems, ethical guidelines and policies that ensure fairness, accessibility and positive emotional and educational impacts (Williamson and Eynon, 2020). This has also been stressed in recent policy publications, such as the US Department of Education (2023), on the importance of advancing equity in AI-assisted education through principles such as human-centred design, oversight and inclusiveness. These priorities are echoed globally, including in OECD and UNESCO guidance [Miao and Holmes, 2023; Organisation for Economic Co-operation and Development (OECD)-Education International, 2023], which advocate for addressing structural inequities when implementing AI in assessment systems. Ethical considerations should involve careful planning to minimise any potential negative consequences on the overall educational experience. Addressing these ethical concerns requires a holistic and thoughtful approach, involving collaboration among educators, policymakers, technologists and other stakeholders to establish guidelines, regulations and best practices for the ethical use of AI in student assessment. However, the existence of multiple guidelines across contexts has created challenges for consistent understanding and implementation. A shared and coordinated perspective is needed to align ethical principles and support coherent policy development for the responsible use of AI in student assessment.

While the advancement of GenAI presents novel opportunities for enhancing learning design and assessment, its integration into student learning also introduces new challenges such as academic integrity and potential deficiency of creativity. Empirical studies have begun to explore the effectiveness of GenAI in educational settings. Early evidence indicates that students using AI-assisted tools demonstrate improvements in learning outcomes, such as enhanced academic writing (Nguyen et al., 2024b; Sun, 2024). However, previous research has also highlighted the risks of over-reliance on GenAI, which may lead to the erosion of students’ skills and knowledge (Gašević et al., 2023). Fan et al. (2025) examined 117 university students and found that the use of AI technologies like ChatGPT may increase learners’ reliance on technology and contribute to what they describe as “metacognitive laziness”, where students engage less in reflective thinking and self-regulated learning. Traditional assessment methods may fall short in capturing the full scope of learning gains achieved through AI-supported learning. This concern echoes recent calls in the assessment literature for more integrative and dynamic approaches that consider both process and performance, including learner-generated evidence and reflective practices (Winstone and Boud, 2022; Bearman et al., 2024).

To address this, researchers have proposed updated assessment frameworks that take into account the unique nature of learning with GenAI (Chan, 2023; Perkins et al., 2024). These frameworks include formative assessments that incorporate AI-generated immediate feedback and summative assessments that focus on evaluating learners’ ability to apply knowledge in novel situations. Additionally, there is an emerging focus on developing assessments for higher-order thinking skills, such as critical thinking and creativity, that can be fostered through AI-driven problem-solving tasks (Giannakos et al., 2024; Nguyen, 2025). Recent studies also emphasise the value of authentic assessment designs that simulate real-world contexts and require learners to apply skills across disciplines, an approach well-aligned with GenAI-supported tasks (Ajjawi et al., 2024).

In line with Carvalho et al. (2022), we argue that not only should learning assessment be reformed to mitigate the risks posed by GenAI, but learning design must also be adapted. Rapid advancements in computing power and AI algorithms, especially with GenAI, are not only influencing education but also driving the automation of key decisions that impact our daily lives and reshape workplaces. GenAI continues to bring about significant change and uncertainty in various sectors, many people may find themselves unprepared for these shifts. A key educational challenge is determining how to equip younger generations with the skills and abilities necessary to adapt to and innovate in an AI-integrated world. To fully leverage the opportunities GenAI presents while minimising its potential downsides for learners, there must be a greater focus on developing higher-order thinking skills. Additionally, recent research highlights the increasing importance of AI literacy in education (Ng et al., 2024; Shalpegin and Nguyen, 2024). This aligns with recent frameworks advocating for data and AI literacies as foundational components of future-ready learning environments [US Department of Education, 2023; Organisation for Economic Co-operation and Development (OECD)-Education International, 2023]. Reforming both learning design and assessment in GenAI-enhanced learning environments is crucial not only for nurturing learners’ development but also for fostering their AI literacy in educational settings.

Despite growing interest in GenAI-assisted learning, there is still uncertainty about how to refine learning design and assessment in this context to maintain academic integrity while preparing the learners for the future workforce (Gašević et al., 2023; Cukurova, 2024; Giannakos et al., 2024; Nguyen, 2025). Various IGOs have issued guidelines for integrating GenAI into education, each offering unique perspectives on its application and challenges. This study aims to synthesise these diverse viewpoints and provide a consensus on a set of guidelines for learning design and assessment in GenAI-integrated learning environments. By consolidating recommendations, the study seeks to offer a cohesive framework for educators and policymakers to effectively leverage GenAI in education.

The increasing integration of GenAI into educational settings necessitates a rethinking of learning design and assessment strategies. In this study, the analysis of guidelines from IGOs provides a comprehensive overview of the skills and competencies that should be emphasised in GenAI-integrated learning outcomes. Documents related to AI in education published prior to the widespread emergence of GenAI, such as UNESCO (2019), were excluded from this analysis. The documents included for analysis are as follows: EC1: the Council of Europe Standing Conference of Ministers of Education on regulating AI in education (Council of Europe, 2023); EC2: The Council of Europe’s recommendations on AI regulation in education (Havinga et al., 2024); EU: the European Commission’s ethical guidelines for educators (European Commission: Directorate General for Education, Youth, Sport and Culture, 2022); OECD1: OECD’s working paper on GenAI in classrooms (OECD, 2023), OECD2: its collaborative policy framework with Education International [Organisation for Economic Co-operation and Development (OECD)-Education International, 2023], UNESCO1: UNESCO’s global recommendation on the ethics of AI (UNESCO, 2021); UNESCO2: guidance for GenAI in education (Miao and Holmes, 2023); and UNESCO3: interpretation of foundation models through its ethics framework (UNESCO, 2023). These guidelines highlight essential factors that must be considered when incorporating GenAI into learning design and assessments for GenAI-enhanced learning, ensuring that they not only evaluate knowledge acquisition but also the learners’ ability to effectively interact with, apply and critically assess AI-driven tools and solutions.

To analyse the policy documents related to learning design and assessment in the context of GenAI, we used a flexible thematic analysis approach (Braun and Clarke, 2006). This method was chosen for its adaptability and robustness in identifying, analysing and reporting patterns within qualitative data, making it particularly suitable for examining the complex and multifaceted nature of policy guidelines. Thematic analysis began with an initial phase of familiarisation with the data. Researchers read and re-read the documents to immerse themselves in the content. During this phase, initial ideas and potential themes were noted. The coding process was iterative and flexible, allowing themes to emerge organically from the data. We used an open coding approach, where data segments were systematically labelled with codes that represented key concepts or issues. This process involved multiple rounds of coding to refine and consolidate codes into coherent themes. We used both deductive and inductive approaches to theme development. Deductively, we applied existing theoretical frameworks and prior research findings to guide the identification of themes. Inductively, we allowed new themes to emerge directly from the data, ensuring that the analysis remained grounded in the actual content of the policy documents.

The IGOs’ guidelines have reached a consensus on the importance of fostering higher-order thinking skills, such as critical thinking, problem-solving and creativity, in learning with GenAI. These skills are deemed essential for navigating the complexities of AI-driven environments. In addition to these cognitive skills, some IGOs have identified digital and AI literacy as emerging competencies that are increasingly in demand in the age of AI. These competencies are crucial for enabling learners to effectively interact with AI technologies, understand their underlying mechanisms and apply them responsibly. Table 1 presents an overview of the key skills and competencies that are required as part of the learning outcomes in AI-integrated educational settings.

The findings indicate a strong emphasis among IGOs on fostering higher-order thinking skills, particularly critical thinking and collaboration/teamwork, suggesting a shared understanding of their importance in modern education. The consistent recognition of these skills reflects their perceived role in preparing learners for complex problem-solving and effective teamwork in a globalised world. The prominent emphasis on critical thinking and collaboration/teamwork among the IGOs aligns with educational literature that identifies these skills as essential for the 21st century. Critical thinking is widely regarded as crucial for enabling learners to analyse information critically and make informed decisions (Pithers and Soden, 2000; Essien et al., 2024). The repeated emphasis across multiple guidelines highlights its perceived importance in fostering analytical abilities necessary for academic and professional success. Similarly, the focus on collaboration/teamwork reflects the growing need for individuals who can work effectively in diverse teams, a skill increasingly important in a connected and collaborative global economy (Järvelä et al., 2023). The alignment with these subthemes suggests that IGOs are responsive to the calls within educational research for curricula that develop these competencies.

Although recognised in documents from organisations like OECD and UNESCO, there is less consensus on subthemes such as flexible thinking, transfer of learning, and metacognition and self-regulated learning, which are highlighted by fewer organisations. This variability may indicate differing priorities or interpretations of these skills’ relevance in educational policies. The variability in emphasis on flexible thinking, transfer of learning, and metacognition and self-regulated learning may reflect differing educational philosophies or stages of integrating these concepts into policy. Flexible thinking and transfer of learning are critical for adaptability and applying knowledge in new contexts (Barak and Levenberg, 2016). Metacognition and self-regulated learning are associated with students’ ability to understand and control their learning processes (Zimmerman, 2002, 2011). The less consistent emphasis might suggest opportunities for IGOs to further incorporate these skills to support lifelong learning. The underrepresentation of metacognition and self-regulated learning is concerning, given their well-established roles in fostering autonomy, resilience and sustained engagement – attributes increasingly necessary in rapidly changing knowledge environments. Without greater integration of these elements, policy guidance may fall short in promoting the full spectrum of competencies needed for learners to navigate complex, dynamic learning ecologies.

In the theme of digital and AI literacy, the emphasis on AI evaluation skills by several organisations highlights an emerging recognition of the need for critical engagement with AI technologies. The recognition of AI evaluation aligns with emerging literature emphasising the importance of critical engagement with AI technologies (Holmes et al., 2021). As AI becomes more integrated into society, the ability to evaluate AI systems critically is essential for understanding their impact and ethical considerations.

The limited emphasis on vocational skills needed to work with AI and AI system auditing suggests a potential gap in addressing the technical competencies required in an AI-driven landscape. There may be insufficient focus on the technical skills required to work with and oversee AI systems. This gap is notable given the increasing demand for AI competencies in the workforce (Green and Lamby, 2023). UNESCO3’s exclusive emphasis on vocational AI skills suggests a recognition of this need, aligning with initiatives that advocate for integrating technical AI education into curricula to prepare students for future job markets. This could point to a focus on general AI literacy over specialised skills in current guidelines. The findings suggest that while IGOs prioritise certain higher-order thinking skills consistently, there is less uniformity in integrating comprehensive digital and AI literacy competencies. This may reflect the rapid evolution of AI technologies and the challenge of updating educational guidelines accordingly. Literature highlights the necessity for education systems to adapt swiftly to technological advancements to equip learners with relevant skills (Luckin and Holmes, 2016; Nguyen, 2025).

Furthermore, the current policy discourse lacks clear, actionable guidance on how AI-related competencies should be taught, to what extent and at what educational levels. Without evidence-based yet concrete and practical approaches, educators may struggle to implement AI literacy in meaningful and pedagogically sound ways, resulting in fragmented or superficial instruction. Similarly, students risk receiving uneven preparation that does not adequately distinguish between basic awareness and more advanced, context-specific competencies required for responsible AI use. The absence of specificity also places a significant interpretive burden on individual institutions and educators, potentially impairing disparities in AI education across regions and systems. To support equitable and effective implementation, there is a pressing need for more detailed and operationalised guidance that addresses both foundational and vocational dimensions of AI literacy, grounded in empirical research and aligned with pedagogical best practices.

Designing GenAI-integrated learning environments requires careful consideration of how GenAI can best support learners by aligning educational objectives, assessment methods and instructional strategies with the capabilities of GenAI. Across international guidelines, several core themes and subthemes emerge that highlight the educational priorities when integrating GenAI. Table 2 presents key factors to consider in designing learning experiences that integrate GenAI. The analysis revealed two main themes: “Rethinking assessment objectives and methods” and “Designing and assessing with AI-inclusive assumptions”. These themes frame the discussion on how GenAI can reshape educational practices to better support learning and assessment.

The findings indicate a strong emphasis among IGOs on rethinking assessment objectives and means, particularly on subthemes like higher-order thinking, cheat-proof tests and formative assessment. This suggests a shared understanding of the need to adapt assessment strategies to better evaluate complex cognitive skills and mitigate academic dishonesty in the context of AI integration.

The consistent recognition of higher-order thinking by OECD1, OECD2 and UNESCO2 reflects the importance placed on developing learners’ critical analysis and problem-solving abilities. The emphasis on rethinking assessment aligns with contemporary educational literature that advocates for assessment methods that better capture students’ complex cognitive abilities and skills relevant to the 21st century. The focus on higher-order thinking corresponds with Bloom’s taxonomy, which highlights the importance of moving beyond rote memorisation to evaluate analysis, synthesis and evaluation skills (Gašević et al., 2023; Nguyen, 2025).

The emphasis on cheat-proof tests by both OECD1 and UNESCO2 highlights concerns about academic integrity in the age of AI, where traditional assessment methods may be vulnerable to misuse. With the advent of AI tools that can generate human-like responses, traditional assessments are increasingly susceptible to cheating (Xia et al., 2024; Fan et al., 2025). This necessitates the development of new assessment strategies that are resilient to such challenges, such as authentic assessments that require personalised and applied responses. However, there is less emphasis on transfer of learning, with alignment from only EC1, OECD1 and UNESCO2. This may indicate differing priorities or challenges in implementing assessments that measure the ability to apply knowledge across contexts.

In the theme of design and assess with AI-included assumptions, only EU, UNESCO2 and UNESCO3 show alignment, focusing on integration in assignments and integration in pedagogical practices, respectively. The limited alignment indicates that integrating AI into assignments and pedagogical practices is still emerging in policy discussions. This suggests that while some IGOs recognise the importance of incorporating AI into educational practices, it is not yet a widespread focus across all organisations. UNESCO’s focus on these subthemes highlights an awareness of the potential of AI to enhance learning experiences when thoughtfully integrated (UNESCO, 2023). Incorporating AI in education can support personalised learning, provide immediate feedback and develop digital competencies (Holmes et al., 2019).

However, the lack of broader alignment among other IGOs suggests challenges in policy adoption, possibly due to concerns about equity, ethical implications or resource constraints. The integration of AI in education requires careful consideration of issues such as data privacy, algorithmic bias and the digital divide (Nguyen et al., 2023). Furthermore, while AI can support personalised learning and cognitive development, current guidelines rarely specify how to design hybrid intelligence systems that balance AI support with the development of transferable skills. Without clearer strategies, there is a risk of undermining core cognitive engagement in complex problem-solving tasks. The findings point to a need for IGOs to expand their guidelines to include strategies for effectively integrating AI into learning design and assessments.

The analysis reveals that IGOs prioritise several key factors in designing assessments for GenAI-integrated learning, aligning with current educational and ethical considerations in the literature. These factors can be categorised into four main themes: GenAI-integrated pedagogical dimensions for assessment, accessibility and inclusiveness, holistic design approach and ethical obligations and principles. These categories provide a structured framework for understanding the essential considerations in designing assessments for learning with GenAI integration. Table 3 presents the essential factors to consider when designing assessments that integrate GenAI.

Under GenAI-integrated pedagogical dimensions for assessment, subthemes such as feedback and support and personalised learning are highlighted by multiple IGOs, suggesting a shared recognition of the importance of adaptive learning experiences in assessments addressing GenAI use in student work. This also aligns with the broader pedagogical shift towards more personalised learning pathways that accommodate each learner’s unique pace, strengths and areas for improvement. The focus on cognitive engagement, specifically higher-order thinking skills such as analysis, evaluation and creation, as well as interaction and dialogue, reflects a broad consensus among organisations on the importance of fostering deep learning. Cognitive engagement is increasingly recognised as essential in preparing learners for complex real-world problems that demand critical thinking and problem-solving skills. AI technologies offer capabilities to facilitate interactive learning environments, where students engage in meaningful dialogues and collaborative activities, enhancing their cognitive engagement and fostering sustained interest in learning tasks (Becerra et al., 2024).

By enabling learners to interact with GenAI-driven feedback and tasks that require advanced cognitive processing, AI-integrated assessments support the development of critical thinking and problem-solving skills, which are core competencies in contemporary educational frameworks. Furthermore, the emphasis on feedback and support within GenAI-integrated assessments is corroborated by studies that illustrate the importance of timely, constructive feedback in bolstering student learning and motivation. GenAI systems can provide immediate, context-specific feedback, guiding students through their learning processes and encouraging self-regulation. Becerra et al. (2024) highlight that personalised feedback enhances learners’ understanding by addressing their individual challenges, thus fostering a supportive learning environment that promotes persistence and reduces the risk of disengagement. This approach aligns with adaptive learning theories, which advocate for instructional support that adjusts to the learner’s current state and promotes incremental progression.

In the theme of accessibility and inclusiveness, several IGOs emphasise technical accessibility. This reflects an awareness of the need to ensure that assessments involving GenAI are accessible to all teachers and learners. It also addresses challenges such as the digital divide and limited access to resources. However, the limited emphasis on AI literacy is noteworthy. Developing AI literacy is essential for enabling learners to understand and critically engage with AI technologies, preparing them for a society where AI is increasingly prevalent (Gašević et al., 2023; Giannakos et al., 2024). The fact that only UNESCO3 emphasises this subtheme suggests that more attention is needed across IGOs to incorporate AI literacy into educational frameworks. Furthermore, there is also a lesser focus on digital education resources and adaptive learning systems, which suggests opportunities for IGOs to further promote inclusive educational technologies that can adjust to diverse learner needs.

From the guidelines, one of the main aspects to consider when designing assessments for GenAI-integrated learning is adopting a holistic design approach. This means developing assessments that not only measure learning outcomes but also account for the broader context of the learning environment (Aqlan et al., 2022), including the interactions between students, AI tools and educators. A holistic design approach focuses on integrating all parts of a system, taking into account its functions as well as the social, cultural and environmental contexts in which it is used. This approach requires designers to look beyond isolated features, instead addressing interdependencies among various components and stakeholders. In learning sciences and educational technology, a holistic design approach is particularly valuable as it supports inclusive, learner-centred environments by acknowledging diverse learner needs, cultural backgrounds and learning contexts (Nguyen et al., 2025). For example, designing a collaborative AI-enhanced learning platform requires balancing technical capabilities with considerations of accessibility, engagement and ethical implications. By fostering a comprehensive perspective, holistic design promotes more resilient and adaptable systems that can evolve alongside learners’ needs and societal changes, thereby supporting sustainable innovation.

Under the holistic design approach, the emphasis on a co-design approach, complementing learner agency and complementing teacher role reflects an understanding of the importance of involving stakeholders in the design process and supporting both learners and educators in AI-integrated environments. The holistic design approach emphasises involving stakeholders through a co-design approach, as recognised by OECD1 and OECD2. This aligns with participatory design principles that advocate for collaborative development processes to create more effective and user-centred technologies (Bødker et al., 2022; Sarmiento and Wise, 2022). Supporting complementing learner agency and complementing teacher role is crucial, as educators and learners are central to the successful integration of AI in education (Gašević et al., 2023; Järvelä et al., 2023). EC1, EU and UNESCO2’s recognition of these subthemes indicates an understanding of the importance of empowering both teachers and students in the learning process. Lastly, the recognition of context and settings by the EC1 and UNESCO3 highlight the need to consider the specific educational environments in which AI is deployed.

In ethical obligation and principles, there is strong alignment on subthemes like data and privacy protection and human-centred principles, indicating that IGOs are cognisant of the ethical considerations inherent in AI integration. The widespread use of GenAI in education (e.g. tools like ChatGPT) raises concerns about issues like plagiarism, the source of training data and the potential for creating and spreading misinformation. Document from EC1, EC2 and EU also highlights the need for ethical frameworks that prioritise transparency, accountability and respect for human rights. This strong emphasis aligns with concerns about the ethical implications of AI in education (Holmes et al., 2021; Nguyen et al., 2023). Ensuring data privacy is critical for protecting learners’ personal information and maintaining trust in AI technologies. Furthermore, the recognition of equity in access and use of AI by OECD2, UNESCO1, UNESCO2 and UNESCO3 highlights the need to address potential disparities in who benefits from AI technologies in education. The emphasis on transparency reflects the necessity to address potential biases and ensure fairness in AI systems. Transparent AI systems allow users to understand how decisions are made, which is essential for accountability and trust. This is particularly important in educational settings, where AI can have significant impacts on learning outcomes and opportunities.

The study set out to systematically examine the policy frameworks guiding the integration of GenAI into learning design and assessment. By analysing policy documents from key IGOs such as the OECD, UNESCO and the European Union, we identified critical themes related to the skills and competencies needed for GenAI-integrated learning, the design of learning experiences and the considerations for assessments in AI-enhanced educational environments.

The research makes several noteworthy contributions to the fields of educational technology and policy development. Luckin and Holmes (2016) advocate for a detailed approach to integrating AI in educational settings, emphasising the importance of aligning technological advancements with pedagogical goals and ethical standards. Similarly, Swiecki et al. (2022) highlight the need for a critical understanding of AI’s role in education, particularly in assessment contexts. In undertaking this study, we draw upon a consensus among policy guidelines on learning design and assessment in the era of GenAI. These perspectives inform our investigation as we examine the policy documents to uncover the frameworks and guidelines that are shaping the future of learning design and assessment, ensuring we use the benefits of GenAI while avoiding potential risks.

Recent studies have consistently highlighted the substantial impact of GenAI on learning assessment, demonstrating its potential to transform how assessments are conducted, adapted and interpreted (Xia et al., 2024). By synthesising guidelines from major IGOs, the study presents an integrated view for learning design and assessment for learning environments incorporating GenAI. Our findings emphasise the imperative to rethink traditional assessment objectives and methods. The focus on developing cheat-proof assessments and integrating GenAI into pedagogical practices reflects the challenges posed by GenAI in maintaining academic integrity and accurately evaluating student learning. This reconceptualisation is crucial for capturing the full scope of learning gains facilitated by GenAI tools. While GenAI presents challenges to academic integrity, it also serves as a catalyst for educational reform, providing new opportunities for adaptive and interactive learning models. Previous studies emphasise that GenAI is assisting educational assessment by offering tools like automated feedback and self-assessment, which foster better learning experiences (Becerra et al., 2024; Messer et al., 2024). This dual role as both disruptor and enhancer highlights the need for AI-aligned policy development to fully leverage its benefits in education.

In addition to expanding integration strategies, our study highlights a critical need for policies to support hybrid intelligence approaches that combine human and AI capabilities in learning environments (Cukurova, 2024; Nguyen et al., 2024b). Such approaches can enhance cognitive development by engaging learners in complex problem-solving tasks where AI serves not as a replacement but as a collaborator that augments human thinking. Yet, without explicit guidance, there is a risk that educational AI use will emphasise efficiency and automation at the expense of deeper learning processes (Fan et al., 2025). Current policies lack sufficient emphasis on how to scaffold the development of transferable skills, such as reasoning, adaptability and self-regulation, when AI tools are embedded in instructional and assessment practices. As learners increasingly interact with AI systems, it is essential that educational design does not offload core cognitive functions to AI in ways that discourage active mental engagement. To ensure that AI integration contributes to long-term cognitive and intellectual growth, robust frameworks and practices must be designed for designing hybrid intelligence systems that preserve and cultivate learners’ higher-order thinking, rather than displacing it.

Furthermore, this study highlights the importance of considering ethical obligations and principles when integrating GenAI into education (Nguyen et al., 2023). Issues such as data privacy, equity in access, transparency and human-centred principles are brought to the forefront in recent IGOs policy guidelines [European Commission: Directorate General for Education, Youth, Sport and Culture, 2022; Havinga et al., 2024; Organisation for Economic Co-operation and Development (OECD)-Education International, 2023; UNESCO, 2021]. This framework serves as a guide for educators and policymakers to navigate the ethical complexities associated with AI technologies in education. Moreover, this study contributes to the discourse on AI implications for educational equity and accessibility, a topic of considerable importance in the field. Prior studies have highlighted the dual aspect of educational technology in both addressing educational equity and increasing digital divides (Nguyen, 2022). In particular with AI, Holmes et al. (2019) highlight the potential of AI to democratise educational opportunities yet also cautions against the risks of exacerbating existing inequalities. Nguyen (2025) further asserts that GenAI tools require an integrated ethical and pedagogical framework to ensure their equitable implementation in higher education, reinforcing the need for policies that balance innovative practices with fairness. As a subset of AI, GenAI also has this dual aspect, which forms a critical part of our policy analysis as we examine how international guidelines address these concerns. Table 4 summarises the implications of our research findings for practice, research and public policy.

The insights from this study have significant implications for various stakeholders. For educators, there is a pressing need to adapt curricula and pedagogical strategies to foster not only cognitive skills but also GenAI literacy. Educators must become proficient in using GenAI tools and in guiding students to use these technologies effectively and ethically. Professional development programmes should be designed to equip teachers with the necessary skills and knowledge. For policymakers, the consolidated guidelines offer a foundation for developing comprehensive policies that address the integration of GenAI in education. Policymakers should prioritise creating equitable access to GenAI technologies, ensuring data privacy and promoting transparency in GenAI applications. Policies must also support the development of infrastructure and resources needed for effective implementation. For researchers, the study opens avenues for further investigation into effective assessment methods in learning environments with GenAI integration. There is a need to explore how higher-order thinking skills can be authentically assessed and how GenAI literacy impacts student outcomes. Researchers should also examine the long-term effects of GenAI on learning and equity.

While the study provides valuable insights, it is limited to policy documents from selected IGOs and may not capture regional or local nuances in policy development. Future research could expand the scope to include a more diverse range of policy documents, including those from developing countries where challenges and opportunities may differ. Additionally, there is a need for empirical studies to test the effectiveness of the proposed guidelines in real-world educational settings, which would provide practical insights into implementation challenges and the impact on student learning outcomes.

In summary, this study provides an overview of guidelines for learning design and assessment for GenAI-integrated learning outlined in the IGOs’ policy documents. It highlights the essential skills and competencies that should be included in the learning outcomes in GenAI-integrated learning. These skills and competencies are categorised into two groups: higher-order thinking skills and digital and AI literacy. It also examines the considerations for designing learning activities and assessments with the integration of GenAI. For learning design, factors that need to be considered include higher-order thinking, transfer of learning, cheat-proof tests, formative assessment, integration in assignments and integration in pedagogical practices. For assessments, considerations focus on four main themes, including GenAI-integrated pedagogical dimensions for assessment, accessibility and inclusiveness, holistic design approach, ethical obligation and principles.

With these insights, this study contributes to a deeper understanding of how GenAI can be responsibly and effectively integrated into learning design and assessment. By highlighting the consensus among major IGOs on essential skills, pedagogical considerations and ethical obligations, it provides a foundational framework for stakeholders to harness the transformative potential of GenAI in education. Embracing these guidelines can help prepare learners for a future where GenAI is ubiquitous, ensuring they are not only consumers of technology but also critical thinkers and responsible innovators. It provides significant implications for teaching and learning practices and policy development on GenAI integration into learning design and assessment, as well as further research on this topic.