Collaborative AI Literacy Framework — SEFI 2025. A systematic review of 9 studies (2015–2023) examining how collaborative learning (CL) approaches can be harnessed to build AI literacy across diverse educational contexts. Using the ICAP framework (Interactive–Constructive–Active–Passive) as an analytical lens, the review demonstrates that CL effectively increases AI literacy across activities, settings, and learner groups. Most studies were conducted in classroom settings, with some broadening participation through educators, families, or AI agents supporting teamwork. Instructional activities spanned all four ICAP modes, revealing a rich design space for collaborative AI literacy interventions.
Authors: (SEFI 2025 Proceedings) · Presented at the 2025 SEFI Annual Conference
Key Findings
This systematic review is among the first to map the intersection of collaborative-learning and ai-literacy, two fields that have largely developed in parallel despite their natural synergies. The review identified 9 studies (2015–2023) that met inclusion criteria for collaborative learning interventions targeting AI literacy outcomes.
The ICAP lens. The authors applied the ICAP framework — which classifies cognitive engagement as Interactive, Constructive, Active, or Passive — to analyze the instructional activities in each study. This revealed that all four modes of cognitive engagement were represented across the corpus, suggesting that AI literacy instruction benefits from a instructional-design approach that sequences activities through multiple engagement levels rather than relying on any single mode. This aligns with broader findings about the effectiveness of active-learning strategies in STEM contexts.
Effectiveness across contexts. Collaborative learning was found to increase AI literacy outcomes across a wide range of instructional activities, educational settings, and learner populations. The studies spanned both formal classroom environments and informal learning contexts, with some interventions deliberately broadening participation — for example, by involving parents and families alongside students, or by incorporating ai-learning-companions-framework AI agents as collaborative partners in the learning process.
Classroom-centric but expanding. Most studies took place in classroom settings, but the review identified emerging patterns of collaboration that extend beyond traditional boundaries: educator-family partnerships, peer-to-peer learning communities, and AI-agent-supported teamwork. These expanding participation models connect to the community-centered-ai-education-adults paradigm and suggest pathways toward more inclusive ai-literacy-equity-programming-policy.
Limited but growing evidence base. The small sample size (9 studies) reflects the nascency of empirical research at this intersection. The authors note the need for more rigorous study designs, larger sample sizes, and longitudinal tracking of AI literacy development — concerns echoed in broader calls for ai-k12-evidence-base research.
Implications
The review's findings have direct implications for instructional-design practice in AI literacy education. The presence of all four ICAP modes across successful interventions suggests that effective AI literacy curricula should intentionally sequence students through passive exposure (e.g., lectures on AI concepts), active manipulation (e.g., hands-on tool use), constructive generation (e.g., creating AI artifacts), and interactive dialogue (e.g., collaborative problem-solving with peers and AI). This multi-modal approach resonates with the ai-literacy-continuum-higher-education framework.
For the collaborative-learning research community, this review provides a structured taxonomy for designing and evaluating AI literacy interventions. The ICAP framework offers a common vocabulary for comparing approaches and identifying which engagement modes are most impactful for different AI literacy competencies — an important step toward building the cumulative science called for in the ai-literacy-assessment-misalignment literature.
Practically, the findings support the integration of collaborative AI literacy activities into existing k-12-ai-education curricula and faculty-development programs. The success of interventions that involve families and community members suggests that AI literacy is not solely a school-based competency but a societal one — a perspective aligned with ai-education-global-capacity and ai-lifelong-learning-policy.
Looking forward, the review highlights the potential for icap-cognitive-engagement-llm-agents research to further enrich collaborative AI literacy instruction by designing LLM-based agents that can serve as interactive learning partners within ICAP-structured activities, extending the social dimension of learning beyond human peers.
Related Pages
- ai-literacy — Core concept: AI literacy definitions and competencies
- collaborative-learning — Collaborative learning theory and practice
- instructional-design — Systematic design of learning experiences
- active-learning — Active learning strategies in education
- icap-cognitive-engagement-llm-agents — ICAP framework applied to LLM-based learning agents
- ai-learning-companions-framework — AI agents as collaborative learning partners
- ai-literacy-continuum-higher-education — AI literacy progression in higher education
- community-centered-ai-education-adults — Community-based AI education approaches
- ai-k12-evidence-base — Evidence base for K-12 AI education
- ai-education-global-capacity — Global capacity building for AI education