AI Literacy

Created: 2026-05-07 | Tags: higher-edk-12policy-makeractive-learningcollaborative-ai-tutoring
📄 Full text: arXiv:2508.15111 · local · Liverpool Repository · local
AI literacy — the knowledge and skills needed to understand, evaluate, and effectively use AI technologies — is increasingly recognized as a core competency. Collaborative learning approaches show consistent effectiveness across activities, settings, and learner groups, and the ICAP framework helps explain why.^hingle-collaborative-ai-literacy-2025

The ICAP Framework

Chi & Wylie (2014) propose that learning activities can be ordered by depth of cognitive engagement:

Mode Student behavior AI literacy example
Passive Receiving information Watching a video about how LLMs work
Active Manipulating information Running prompt experiments, observing outputs
Constructive Generating new ideas Designing a prompt engineering task for peers
Interactive Co-constructing with others Debating AI ethics in a group, building a shared policy

Hingle & Johri's systematic review (9 studies, 2015–2023) found that collaborative learning activities for AI literacy included all four ICAP modes. This suggests that effective AI literacy programs should span the full engagement spectrum, not just information delivery.

Collaborative Learning for AI Literacy

Settings and Participants

- Educators and families (intergenerational AI literacy) - AI agents supporting teamwork (human-AI collaborative learning)

Key Insight

Collaborative learning was effective across diverse contexts, but the mechanism was not uniform collaboration per se — it was the ICAP depth of the collaborative activity. Interactive (co-constructive) activities produced stronger literacy outcomes than Active or Passive ones.

Relationship to AI-in-Ed Research

AI literacy application Wiki connection
Understanding how tutoring AI works tutoring-specific-vs-general-ai — students should know whether they're interacting with a general LLM or pedagogically optimized system
Evaluating AI feedback quality ai-peer-feedback-systems, formative-assessment — literacy to assess whether AI-generated feedback is useful
Prompt engineering as constructive activity pedagogical-llm-training — reverse-engineering effective tutoring behavior
AI ethics and policy co-construction faculty-development-genai, authentic-assessment — institutional literacy for governance

Multimodal AI Literacy Levels (Varga-Atkins et al., 2025)

The Educators' Guide distinguishes three operational levels of multimodal GenAI literacy:

Level Capabilities Example Activities
Basic Awareness of multimodal GenAI platforms, capabilities, and appropriate uses Creating prompts; generating visual outputs; understanding platform limitations
Intermediate Co-create multimodal content; critically evaluate outputs; scaffold uses for peers Transforming lecture notes into visuals or podcasts; annotating AI-generated essays or images for bias
Advanced Design activities/assessments incorporating multimodal GenAI; lead ethical/philosophical discussions Redesigning a module to include GenAI co-creation; institutional policy drafting; graduate attribute alignment

These levels operate across four scales:

This framework complements the ICAP engagement hierarchy by adding institutional scaling and multimodal specificity. Where ICAP asks how deeply students engage, the multimodal literacy framework asks at what level and at what scale.

Implications for Curriculum Design

1. Start passive, move to interactive — Begin with conceptual foundations, progress to co-construction 2. Use AI agents as collaborative partners — Not just tools to learn about, but teammates to learn with 3. Embed assessment in co-construction — AI literacy assessment can itself be an interactive activity (e.g., group policy debates) 4. Foster critical evaluation — Students must be able to detect pedagogical harms, bias, and misinformation in AI outputs

Open Questions

1. Does AI literacy transfer across domains (e.g., from chatbot evaluation to algorithmic bias detection)? 2. How does AI literacy interact with SRL — does knowing how AI works change self-regulation strategies? 3. Can affective AI support or undermine AI literacy development?

Related Pages

📎 107 other pages tagged ai-literacy