Co-designing AI-enhanced careers education: a design-based approach to sustainable career ecosystems

The accelerating influence of artificial intelligence (AI) is transforming how individuals learn, work, and plan for the future. As automation, datafication, and intelligent systems reshape the global labour market, educators face growing pressure to prepare students for careers that do not yet exist. Schools and universities must move beyond static career guidance and towards adaptive, ethically grounded learning ecosystems that promote lifelong adaptability, reflective agency, and critical engagement with technology. This perspective is developed in detail through the SAICEM framework introduced later in the paper.

This study examines how AI can be strategically and ethically integrated into the design of educational interventions that support students' transitions to future work. While the initial context for this research was secondary career learning, the paper's broader objective is to explore AI as a catalyst for reimagining the design process itself, how educators, industry partners, and researchers collaboratively conceptualise, govern, and implement AI-enabled learning environments. The study reframes AI not as a replacement for human expertise, but as a co-participant in the pedagogical ecosystem that can extend teacher capacity, foster personalisation, and illuminate ethical and social tensions inherent in technological change.

Guided by a design-based research (DBR) methodology, the study engaged teachers, a national manufacturing industry body, and academic experts in education and AI to collaboratively design a short course addressing skills, ethics, and decision-making for work in AI-mediated contexts. The process produced the Sustainable AI Career Ecosystem Model (SAICEM), a conceptual framework positioning AI as both artefact and actor within educational ecosystems. SAICEM integrates governance, ethics, and pedagogy as interdependent dimensions of responsible AI design and offers a foundation for future research and practice across diverse educational settings.

This paper reports specifically on Cycle 1 (co-design) of the broader DBR project, which focused on the participatory development of the conceptual model and design principles underpinning SAICEM. Subsequent cycles will evaluate and refine the model through implementation and empirical validation. By documenting this first phase, the paper contributes to the growing body of research exploring how AI can be meaningfully embedded within participatory design processes to create equitable, transparent, and sustainable educational innovations.

The integration of artificial intelligence (AI) into education intersects with multiple research traditions, including career development, educational design, and socio-technical studies of technology in learning. This review positions the study within three interrelated bodies of literature: (1) career guidance and education theory, (2) AI in education and learning design, and (3) ecosystemic approaches to sustainability and equity.

Career education has evolved from prescriptive vocational guidance to learner-centred approaches that emphasise agency, adaptability, and meaning-making (Hooley et al., 2017). Foundational theories such as Social Cognitive Career Theory (Lent et al., 1994) highlight the importance of self-efficacy, outcome expectations, and goal-setting in shaping career decision-making. Similarly, Career Construction Theory (Savickas, 2005) conceptualises career development as a narrative process through which individuals build identity and purpose in response to social and economic change. Contemporary frameworks increasingly integrate these perspectives with systemic and ecological approaches that account for the influence of schools, communities, and policy environments (Watts and Fretwell, 2004; Kettunen and Sampson, 2019).

Recent scholarship has emphasised the need to reconceptualise career learning as a process of developing career adaptability and career capital, the accumulation of skills, social networks, and reflective competencies necessary to navigate fluid labour markets (Blokker et al., 2023; Donald and Jackson, 2023). However, most school-based programmes continue to focus on individual guidance activities rather than systemic coordination or ethical engagement with emerging technologies. This study responds to this gap by exploring how generative AI can extend traditional career education models towards participatory and ethically aware educational ecosystems.

AI technologies have introduced powerful affordances for personalisation, simulation, and data-driven feedback in education. Large language models (LLMs), chatbots, and generative AI tools enable learners to engage in interactive dialogues, explore simulated workplaces, and receive adaptive feedback (Luckin and Holmes, 2016; Holmes et al., 2019). Within career contexts, AI can emulate the functions of guidance counsellors, assist with labour market exploration, and support students in articulating skills and pathways (Vicsek, 2021). Nevertheless, critical scholars caution against uncritical adoption. Automation bias, the erosion of teacher autonomy, and the amplification of algorithmic bias risks reinforcing inequities and reducing human oversight (Watters, 2023; Erstad et al., 2023).

Educational design research highlights the need to treat AI not merely as a technological enhancement but as an actor that co-shapes pedagogical relationships (Selwyn, 2019; Popenici and Kerr, 2017). Frameworks such as Fawns' (2022) entangled pedagogy underscore the inseparability of technology, pedagogy, and context, calling for participatory design processes that foreground ethics and human agency. These perspectives inform this study's methodological commitment to co-design, ensuring that AI integration aligns with teacher expertise, curriculum legitimacy, and inclusive governance.

The concept of sustainability in career development has expanded beyond employability to encompass wellbeing, equity, and long-term adaptability (Donald et al., 2024). Career ecosystems (Baruch, 2015) describe interdependent systems connecting individuals, institutions, and labour markets, while the more recent Sustainable Career Ecosystem model (Donald et al., 2024) highlights the dynamic interplay between individual agency and institutional structures. Yet, these models often treat technology as an external factor rather than an embedded participant in shaping career trajectories.

The Sustainable AI Career Ecosystem Model (SAICEM) proposed in this study extends this literature by conceptualising AI as both an artefact and actor within educational ecosystems. By integrating governance, ethics, and pedagogy within a single framework, SAICEM extends existing ecosystem models that typically treat these domains separately, enabling a more coherent account of how educational, institutional, and moral considerations intersect in AI-enabled contexts.

SAICEM integrates principles of sustainability, ethical governance, and participatory pedagogy, responding to calls for more systemic and inclusive approaches to technology integration (Percy and Hooley, 2024). It positions the design of AI-supported education not as a technical implementation challenge, but as an ethical and collaborative process that must reconcile innovation with equity and professional autonomy.

Bringing these strands together, the literature reveals several key insights:

• Career education must evolve from individualistic guidance models to systemic, adaptive, and ethically aware frameworks.

• AI offers new opportunities for personalisation and reflection but demands critical governance and participatory oversight.

• Design-based research provides a methodological bridge between theory and practice, enabling iterative co-creation of AI-enhanced educational models.

This synthesis positions SAICEM as both a product and process of collaboration, linking career learning theory, ethical AI design, and educational sustainability. The literature thus establishes the rationale for using DBR to investigate how AI can be meaningfully embedded in educational ecosystems to support equitable and future-ready learning.

This study employed a design-based research (DBR) methodology to explore how artificial intelligence (AI) can be ethically and effectively integrated into the design of educational programmes that prepare students for sustainable and adaptive futures. DBR was selected for its capacity to bridge theory and practice through iterative, collaborative cycles of analysis, design, implementation, and evaluation (Design-Based Research Collective, 2003). It is particularly suited to investigating complex educational problems situated within authentic contexts and involving diverse stakeholders. The approach enables the generation of both practical solutions and theoretical insights through continuous engagement with real-world design processes.

The overarching DBR project comprises three iterative cycles conducted over an 18-month period:

• Cycle 1 (Co-design): Development of a conceptual framework and design principles.

• Cycle 2 (Implementation): Pilot testing of AI-enabled learning modules in secondary schools.

• Cycle 3 (Evaluation): Refinement of the Sustainable AI Career Ecosystem Model (SAICEM) through empirical analysis.

Reporting exclusively on Cycle 1 (Co-design), the purpose of this phase was to engage educators, industry partners, and academic experts in collaboratively conceptualising the role of AI in educational innovation and developing the foundational principles of SAICEM. Implementation and evaluation activities are beyond the present scope and will be addressed in subsequent publications.

Cycle 1 involved three principal stakeholder groups:

• Teachers – experienced career education practitioners contributing pedagogical and curricular expertise.

• Industry Partners – representatives from a national manufacturing industry body providing insights into emerging skill needs and workplace realities.

• Academic Experts – researchers in education, AI, and learning sciences contributing theoretical and ethical perspectives.

This configuration reflects a meso-level convergence between schools, industry, and higher education, consistent with sustainable career ecosystem principles (Donald et al., 2024). Stakeholder diversity was critical to ensuring that the co-design process captured pedagogical, ethical, and contextual nuances while maintaining curriculum legitimacy.

Following participatory design principles (Sanders and Stappers, 2008), the research team conducted a series of structured workshops over four months. Each session employed collaborative activities to identify skill gaps, map curriculum alignment opportunities, and explore the responsible use of generative AI tools such as ChatGPT, HeyGen, and LMNT for dialogue, video, and voice synthesis.

The workshops were designed to facilitate mutual learning rather than consultation, positioning all participants as co-creators. Teachers contributed contextual insights regarding curriculum integration and student engagement; industry partners articulated authentic workforce challenges; and academics ensured ethical oversight and theoretical coherence. This triadic interaction supported the development of design tensions and propositions that informed the creation of SAICEM.

While student participants were not included in Cycle 1, their involvement is planned for Cycles 2 and 3 through pilot testing, feedback sessions, and embedded student co-design activities. This staged inclusion reflects both ethical caution and the iterative structure of DBR, ensuring that initial prototypes are refined before being tested with learners.

Data collection in Cycle 1 focused on capturing stakeholder dialogue and collaborative artefacts generated during workshops. Sources included:

• Facilitator notes and reflective memos;

• Audio recordings and transcriptions of co-design sessions;

• Artefacts such as curriculum maps, ethical decision matrices, and prototype AI prompts; and

• Post-workshop reflections from participants.

A thematic analysis approach (Braun and Clarke, 2006) was used to identify key themes across the dataset, focussing on stakeholder values, ethical concerns, pedagogical design considerations, and perceptions of AI's role in education. The analysis informed the development of SAICEM's triadic structure, governance, ethics, and pedagogy, linking theoretical constructs with practical design implications.

Ethical approval was granted by the university's Human Research Ethics Committee (Protocol #2023/45). Participation was voluntary, with informed consent obtained from all contributors. Data were anonymised and stored securely in accordance with institutional data management policies. Special attention was given to power relations within co-design workshops to ensure balanced participation among educators, industry representatives, and academics. Transparency regarding AI tool use, data privacy, and authorship of co-created content was maintained throughout.

Cycle 1 produced two principal outcomes:

• A pilot-ready curriculum framework for an AI-enhanced short course designed to support students' exploration of emerging career pathways; and

• The conceptualisation of the Sustainable AI Career Ecosystem Model (SAICEM) synthesises design tensions and stakeholder perspectives into a structured framework for ethical AI integration in education.

SAICEM emerged from the iterative synthesis of stakeholder dialogue and theoretical reflection. It positions AI as both artefact and actor within the educational ecosystem, mediating interactions among learners, teachers, and institutional structures. The model's three dimensions (governance, ethics, and pedagogy) provide a scaffold for future implementation and evaluation cycles, aligning educational innovation with principles of inclusivity, transparency, and sustainability.

The findings from Cycle 1 of the design-based research (DBR) project illustrate how collaborative engagement among educators, industry partners, and academic researchers produced both tangible and conceptual outcomes. These outcomes advance understanding of AI integration in educational design and culminate in the Sustainable AI Career Ecosystem Model (SAICEM), presented in Figure 1.

The participatory workshops revealed the complexity of embedding AI within school-based learning environments. While stakeholders recognised the pedagogical potential of generative AI for personalisation, scenario-based learning, and reflective dialogue, they also expressed concerns regarding ethical transparency, teacher workload, and curriculum validity. Teachers, in particular, emphasised the importance of maintaining professional agency and curricular integrity when adopting AI-driven tools.

While teachers emphasised concerns around curriculum legitimacy, industry partners focused more strongly on future skills alignment, revealing productive tensions that informed the final model.

Three key design tensions emerged:

• Automation vs Authorship: Balancing the efficiency of AI-generated guidance with the need for authentic student reflection and teacher moderation.

• Innovation vs Legitimacy: Aligning emerging AI applications with curriculum standards and assessment requirements to ensure institutional acceptance.

• Access vs Equity: Addressing disparities in digital infrastructure and ensuring inclusive participation for schools in low-resourced settings.

These tensions informed the development of design propositions that underpin SAICEM, ensuring the framework is grounded in practical realities rather than purely theoretical constructs.

The collaborative synthesis of stakeholder insights and theoretical perspectives culminated in the development of the Sustainable AI Career Ecosystem Model (SAICEM). By integrating governance, ethics, and pedagogy within a single framework, SAICEM extends existing ecosystem models that typically treat these domains separately, enabling a more coherent account of how educational, institutional, and moral considerations intersect in AI-enabled contexts. Figure 1 presents the model's triadic structure, positioning AI as both a technological artefact and a socio-technical actor within sustainable learning ecosystems.

This conceptualisation extends existing understandings of AI in education by reframing it as a socio-technical participant rather than a neutral tool, aligning with emerging work on relational and entangled pedagogies. Its three interrelated domains, governance, ethics, and pedagogy, form a dynamic framework for co-designing and implementing AI-enabled learning environments that uphold transparency, professional agency, and inclusivity. The figure visualises the conceptual outcome of DBR Cycle 1 and provides the foundation for subsequent implementation and evaluation phases.

Within SAICEM:

• Governance addresses accountability, data ethics, and institutional transparency in AI decision-making.

• Ethics encompasses fairness, inclusivity, and respect for human autonomy in AI-enhanced pedagogy.

• Pedagogy involves integrating AI to support learner agency, metacognition, and critical reflection rather than replacing teacher expertise.

This triadic configuration positions AI not merely as a technical aid but as an embedded agent mediating relationships between human participants and institutional systems. By visualising AI's participatory role, SAICEM offers a framework for developing ethically grounded educational designs that are sensitive to social and contextual factors.

To situate SAICEM within existing theoretical landscapes, it was compared with prior frameworks in vocational education, sustainable career development, and AI in education. Figure 2 summarises this comparison, highlighting SAICEM's distinct contribution to integrating socio-technical perspectives into educational design.

The table contrasts SAICEM with prior conceptual models, including organisational and sustainable career ecosystems (Baruch, 2015; Donald et al., 2024) and AI-in-education frameworks (Luckin and Holmes, 2016; Holmes et al., 2019). SAICEM uniquely positions AI as an active socio-technical participant rather than an external tool, foregrounding inclusive governance, ethical accountability, and pedagogical alignment across multiple stakeholder levels.

This comparative synthesis clarifies SAICEM's theoretical innovation. While traditional career ecosystem models focus on employability and institutional coordination, and AI-education models emphasise adaptive learning, SAICEM merges these perspectives into an integrated framework for designing sustainable, human–AI partnerships in education.

The findings reinforce the value of DBR as a methodological approach for developing contextually grounded AI integration strategies. The iterative co-design process surfaced design principles that will inform Cycles 2 and 3:

• Prioritise ethical deliberation alongside technical innovation in every design iteration.

• Embed teacher agency and student voice as non-negotiable design requirements.

• Ensure institutional and policy alignment to legitimise AI adoption in curriculum reform.

• Use generative AI not as a replacement for human expertise but as a reflective partner in learning.

These insights align with and extend current discussions in the educational technology field regarding the responsible adoption of generative AI. By articulating SAICEM as both an artefact and methodological tool, the study contributes to a growing body of scholarship concerned with the socio-ethical design of AI systems in education.

Cycle 1 of the DBR project achieved three key contributions:

• It produced SAICEM, a conceptual model that redefines AI's pedagogical and institutional role within education.

• It generated design principles for ethical and sustainable AI integration, grounded in stakeholder collaboration.

• It established a methodological precedent for combining co-design and ecosystemic frameworks to investigate complex educational innovations.

Collectively, these findings provide a foundation for the subsequent cycles of implementation and evaluation, which will empirically validate SAICEM's theoretical claims and assess its adaptability across educational sectors.

The two conceptual figures presented in this paper synthesise the key outcomes of Cycle 1. Figure 1 visualises the Sustainable AI Career Ecosystem Model (SAICEM), illustrating how governance, ethics, and pedagogy interact as dynamic dimensions within participatory AI-enabled educational design. Figure 2 situates SAICEM in relation to existing ecosystem and AI-in-education frameworks, clarifying its distinctive contribution as a socio-technical, ethically grounded model. Together, these figures encapsulate the theoretical and practical advances of this study, demonstrating how design-based research can yield not only contextual innovations but also transferable frameworks for future-ready and inclusive education.

This study examined the role of artificial intelligence (AI) as both a catalyst and collaborator in the participatory design of educational innovations. Grounded in a design-based research (DBR) methodology, the first cycle (Cycle 1) documented the co-design process undertaken with teachers, industry representatives, and academic experts to conceptualise AI's ethical and pedagogical integration within secondary education. This process resulted in the development of the Sustainable AI Career Ecosystem Model (SAICEM), as a means of demonstrating the implications of treating AI as an active participant in educational ecosystems, particularly for governance, professional agency, and ethical responsibility.

The findings of this first DBR cycle demonstrate that when AI is integrated through a structured, participatory process, it can enhance stakeholder collaboration, support personalised learning design, and surface important tensions between innovation and professional agency. SAICEM responds to these tensions by framing AI as a socio-technical actor rather than a neutral instrument, an actor that co-shapes decision-making, equity, and ethical responsibility within educational systems. This reconceptualisation challenges deterministic narratives of automation by emphasising teacher capacity, learner agency, and institutional transparency.

The theoretical contribution of this paper lies in extending design-based research toward an ecosystemic understanding of AI in education. SAICEM provides a conceptual bridge between educational design and ethical AI governance, illustrating how participatory processes can produce frameworks that are both practically applicable and theoretically generative. While the initial design context focused on career learning within the manufacturing sector, the model's principles are adaptable to broader educational domains. Its triadic structure, governance, ethics, and pedagogy, offers a replicable foundation for designing AI-enabled learning ecosystems that uphold inclusivity and sustainability.

From a practical perspective, SAICEM supports educators, policymakers, and industry partners in co-developing AI-driven curricula that balance innovation with accountability. It aligns with global priorities such as the United Nations Sustainable Development Goals, particularly SDG 4 (Quality Education) and SDG 8 (Decent Work and Economic Growth). By embedding human oversight, ethical review mechanisms, and contextual sensitivity, the model fosters trust in AI-enabled learning environments.

Future research will extend this work into Cycles 2 and 3 of the DBR project, involving pilot implementation, student participation, and iterative evaluation of SAICEM's effectiveness across diverse educational sectors. These next phases will empirically validate the model's propositions, refine its components, and explore its transferability beyond the initial context. Through this continuing work, SAICEM aims to contribute to the development of equitable, transparent, and sustainable frameworks for integrating AI into educational practice, positioning AI not as a replacement for educators, but as a partner in the shared pursuit of meaningful and adaptive learning futures.