Students' Perceptions of Multiliteracies Development Using AI-Assisted Portfolio Assessment

Synthesis: Ni and Lam (2026) report a 16-week qualitative study in which 50 first-year Tourism Management students at a Chinese university collaboratively compiled WordPress-based e-portfolios of Multimodal AI writing about touring-route recommendations, supported by Generative AI (GenAI) tools as direct Feedback providers and as moderators/synthesizers of peer and teacher feedback. Across reflective journals, focus-group interviews, and narrative inquiries, students developed multiliteracies through three stages, transforming from passive to proactive learners while gaining multimodal awareness, confidence, metacognition, and professional readiness. Students perceived GenAI as a study companion offering instant, gentle, rubric-referenced Feedback, but also faced challenges including over-trust in AI and information overload. The study positions GenAI-assisted portfolio assessment as a transformative literacy practice within Formative Assessment.

Key Findings

• Three-stage multiliteracies trajectory: students moved from AI-empowered curation and redesign of multimodal artifacts, to augmented confidence in multimodal meaning-making, toward readiness for real-world multimodal communication — a shift from passive recipients of knowledge to proactive, self-regulated learners.
• AI-enhanced Student Engagement: the recursive create–curate–reflect–revise cycle, scaffolded by GenAI suggestions on imagery, color schemes, and layout, deepened engagement and design thinking (e.g. Ella's group adopting low-saturation colors; Student 12 harmonizing a color palette and adding a Li Bai quotation hook).
• Expanded usefulness of portfolios: students gained competence in managing digital platforms (WordPress), crafted audience-aware artifacts with a global reach, and connected the project to career demands (CET certificates, video editing, multimodal self-media aesthetics).
• Critical multisource feedback evaluation: students learned to selectively integrate peer, teacher, and AI feedback rather than adopt it wholesale, using AI to explain or synthesize teacher comments — though many initially treated AI output as authoritative (Lucy: "I'd unconsciously trust it").
• Metacognitive and confidence gains: gradeless, feedback-rich portfolio compilation (vs. high-stakes exams) reduced writing anxiety, fostered learner agency, and surfaced "aha moments" where self-assessment diverged from teacher/AI feedback (e.g. FG3-S1 revising "red spirit" to "revolutionary spirit" for intercultural clarity).
• Co-constructed rubrics as catalysts: rubrics jointly built with students guided self- and peer-assessment, feedback provision, and better prompting of AI for rubric-referenced feedback.

Study Design & Method

• Design: qualitative, interpretivist study over a 16-week semester examining students' multiliteracies development and perceptions of GenAI use in portfolio compilation; guided by two research questions on development and on perceived affordances/challenges.
• Participants: 50 first-year Tourism Management majors (CEFR A1–B2) with no prior academic AI experience, in an application-oriented Chinese university.
• Intervention: students worked in 11 groups to create WordPress-based individual expository + argumentative essays synthesized into a PowerPoint group portfolio on touring-route recommendations; AI feedback (e.g. ERNIE-BOT, Kimi) provided immediate feedback on language, organization, and multimodal integration, and facilitated feedback uptake and synthesis across three weekly writing cycles.
• Data & analysis: 50 reflective journals, 4 focus-group interviews (~1 hour each), 6 narrative inquiries (30–45 min each), plus multimodal analysis of artifacts; iterative inductive thematic analysis conducted in Chinese and translated to English by the author team.

What this means for practice

• Instructors. Pair GenAI-supported feedback with gradeless, feedback-rich portfolio cycles rather than high-stakes examinations; in this 16-week study the low-stakes multisource format reduced writing anxiety and increased engagement while students kept agency over their revisions.
• Instructors. Co-construct the rubric with students and extend it beyond linguistic accuracy to multimodal orchestration — imagery, color scheme, layout, audience — so that rubric-referenced prompts return usable AI feedback for self- and peer-assessment.
• Instructors. Scaffold critical AI engagement explicitly. Without instruction, students treated AI output as authoritative ("I'd unconsciously trust it"); teach prompt refinement, comparison across teacher, peer, and AI feedback, and criteria for selective uptake.
• Designers. Position GenAI as a co-partner and feedback synthesizer rather than a replacement for teacher or peer comment, since students gained most when AI explained or integrated others' feedback while they retained revision decisions.
• Administrators. Provide structured support and training so GenAI-assisted formative assessment stays equitable and does not displace learner autonomy or critical thinking.

Limitations

• The sample is 50 first-year Tourism Management majors at one Chinese university, all recently through the College Entrance Examination with no prior academic AI experience; the authors note the effects of the AI workshop and guidance were therefore "salient" and call for testing in senior-year classes with different AI-literacy levels.
• The design is qualitative and self-report (50 reflective journals, 4 focus-group interviews, 6 narrative inquiries) with no comparison or control condition, so it documents perceptions rather than measured multiliteracies gains.
• GenAI was used almost entirely for dialogic feedback and assessment, with little application in multimodal image creation or visual design, so the creative dimensions of the claimed multiliteracies development rest on thin evidence.
• Data were collected in a single 16-week course, analyzed in Chinese, and translated into English by the author team, adding an interpretation layer to already context-bound findings.

Citation

Ni, Y., & Lam, R. (2026). Students' perceptions of multiliteracies development using AI-assisted portfolio assessment. Pedagogies: An International Journal, 21(2), 301–328.