Fostering responsible AI literacy: A systematic review of K-12 AI ethics education

Faculty of education and partner education university department

Hong Kong SAR, China

Systematic search, screening, synthesis, framework proposal

• In-school (K–12)

PRISMA-style systematic review of empirical AI ethics education literature

Corpus Jan 2014–Mar 2025 (68 publications)

Multi-study synthesis

Classroom implementations vary; includes discussions of LLM-era ethics

• Ethics classroom practice lags policy rhetoric globally
• Assessment of ethical learning is methodologically difficult
• East/West context differences complicate transfer
• Emerging GenAI ethics evolve faster than publication cycle

K-12 across included studies

Increasing exposure to GenAI and school AI tools

Varies; ethics focus can be decoupled from advanced coding

• Map how K-12 AI ethics is taught and assessed internationally
• Synthesize outcomes across cognitive, affective, and behavioral domains
• Propose competency-based responsible AI literacy framework integrated across AI literacy dimensions

• Highlight implementation gaps (e.g., minimal instructional time on ethics in some curricula)
• Guide educators and policymakers on actionable responsible-AI priorities

• Not a trial of one ethics curriculum
• Not exhaustive of non-peer-reviewed practice guides
• Not solely technical AI skill training

Pedagogical approaches spanning discussions, projects, and critical examinations of AI systems (corpus-dependent)

• Evaluator

Global literature; East/West contrasts noted

• Classroom discourse and structured ethical inquiry
• Activities examining bias, privacy, opacity, manipulation, hallucinations, copyright, creativity impacts

• Use developmentally appropriate cases for sensitive topics
• Avoid performative ethics tick-boxing without behavioral follow-through
• Protect students discussing surveillance or injustice without retraumatization
• Update materials as LLM risks evolve

• Systematic search and screening to K-12 AI ethics empirical education studies
• Extract pedagogies, RAI principles, assessments, outcomes
• Compare contexts and identify gaps
• Derive competency-based responsible AI literacy framework and research agenda

• Students learn to scrutinize AI sociotechnical impacts; institutions must supply time, materials, and PD

• Integrate ethics across technical AI learning dimensions rather than siloed add-on

• Large gap between recognition of AI ethics in frameworks and classroom time devoted to it
• Assessment of ethical learning remains limited and inconsistent
• Emerging GenAI ethics not uniformly addressed

• Reconceptualizes AI ethics as transformative dimension permeating all AI literacy dimensions
• Connects synthesis to UNESCO-style curriculum-time observations in discussion

• Catalogs diverse pedagogical designs and assessment approaches in the field

• Shows ethical learning outcomes manifest across cognitive, affective, and behavioral domains where measured

Concludes with three future research directions to strengthen K-12 responsible AI literacy evidence and practice.

• Responsible handling of student discourse on bias, justice, and surveillance
• Academic integrity shifts with GenAI require updated classroom ethics
• Equity across contexts (e.g., differing policy enforcement East/West)
• Transparent criteria when assessing ethical reasoning without punitive high-stakes misuse

• Activity documentation
• Practitioner observation

• Develop validated ethical reasoning rubrics for K-12
• Longitudinal studies linking ethics instruction to behavior with AI tools
• Cross-cultural comparative implementation trials

• AI ethics education
• K-12 digital citizenship
• Responsible AI / RAI competencies

• Completed

K-12

Formal schooling (global corpus)

Ethics / responsible use across literacy dimensions

Integrated ethics inquiry

Medium

Medium