Feasibility of implementing a multicultural curriculum through artificial intelligence: perspectives of educational science experts

As artificial intelligence (AI) tools increasingly permeate education systems, they intersect with long-standing efforts to design and implement multicultural curricula aimed at equity, representation, and critical consciousness. Policy frameworks emphasise “human-centred” AI and cultural diversity. Yet little is known about how educational science experts assess the feasibility of using AI to advance multicultural curriculum reforms rather than undermine them, particularly in context-specific higher-education settings such as China, where digital transformation and curricular governance shape the conditions for implementation. This study investigates how experts in educational sciences conceptualise the feasibility of implementing a multicultural curriculum through AI and, under what ethical, epistemic, pedagogical, and structural conditions, such implementation is perceived as desirable and realistic in the context of Chinese higher education. Adopting an interpretivist, qualitative design, the study drew on semi-structured interviews with purposively selected educational science experts from the faculties of curriculum studies, educational technology, comparative education, philosophy of education, and teacher education. Inclusion criteria required a doctoral degree, at least five years of post-doctoral experience, and demonstrable engagement with multicultural and/or technology-enhanced education. Interviews (60–90 min) were audio-recorded, transcribed verbatim, and analysed using reflexive thematic analysis supported by qualitative data analysis software. Trustworthiness was enhanced through iterative coding, analytic memoing, reflexive journaling, and limited member checking. Participants were recruited from two Chinese higher-education institutions, and feasibility was treated as a context-bound educational judgment rather than a universal property of AI tools. Analysis initially generated nine lower-order categories, which were subsequently synthesized into four higher-order themes. Experts framed AI as a conditional enabler of multicultural differentiation, capable of diversifying languages, examples, and representations when anchored in explicit multicultural aims. Simultaneously, they warned of risks of epistemic injustice and cultural homogenisation arising from biased data and opaque algorithms. Critical AI literacy and reconfigured teacher roles as “curriculum mediators” were seen as essential. Feasibility was strongly linked to structural and policy conditions, assessment and accountability regimes, contextual variation across systems, the need to balance innovation with caution, and requirements for interdisciplinary collaboration. Experts also highlighted students’ role as critical co-constructors who interrogate, rather than consume, AI outputs. Across the revised thematic structure, feasibility was constructed through four interrelated domains: conditional pedagogical feasibility, epistemic and ethical risk, human mediation and institutional governance, and democratic co-construction of multicultural learning. Educational science experts regard AI-mediated multicultural curricula as possible but structurally fragile. Feasibility depends on value-led curriculum design, robust governance, and critical human mediation that aligns AI with the transformative ambitions of multicultural education. More specifically, feasibility depends on the alignment of explicit multicultural curricular intent, recognition of epistemic and ethical risk, strong institutional and pedagogical mediation, and students’ active participation in critical inquiry.