Supporting Holistic AI Ethics Literacy Education Through Critical Reflection: Three Recommendations for Fostering Children’s Ethical Growth

Paper Title

Supporting Holistic AI Ethics Literacy Education Through Critical Reflection: Three Recommendations for Fostering Children's Ethical Growth

Publication Info

• Topic area: AI ethics literacy education for children
• Keywords: AI ethics, children, critical reflection, design fiction, UNESCO principles, scaffolding, computational empowerment, ethical reasoning, speculative methods, HCI

Background and Problem

• Problem / challenge: Despite the increasing presence of AI in children's lives, foundational diagnostic knowledge about their ethical literacy is lacking. Current efforts often address isolated ethical topics without a holistic structure, leaving gaps in understanding how children integrate multiple ethical principles.
• Significance: Addressing this gap is critical to mitigate potential harms, biases, and misunderstandings arising from AI interactions, and to foster children's digital literacy, agency, and critical engagement with AI.
• Motivation and related work: Prior studies have shown that children do not naturally develop a scientific understanding of AI through usage alone. While interactive and narrative methods have demonstrated promise in teaching ethical concepts, existing approaches often focus on stand-alone issues or specific contexts, neglecting broader ethical integration.

Solution

• Proposed approach: A design fiction-based method using speculative scenarios aligned with UNESCO’s ethical principles for AI to foster critical reflection and ethical reasoning in children aged 10-11.
• Novelty:
  1. Empirical mapping of children's ethical reflections across multiple AI principles.
  2. Evidence that design fiction scenarios can induce measured caution rather than outright rejection of AI.
  3. Three actionable recommendations for holistic AI ethics literacy education.
• Procedure and key techniques:
  • Co-design phase involving educators, psychologists, and AI experts to create 10 speculative scenarios based on UNESCO principles.
  • Classroom-based study with 66 children using semi-structured group discussions, pre/post-surveys, and thematic coding of qualitative data.
  • Mixed-methods analysis to identify patterns in ethical reasoning and attitudinal shifts.

Results

• Concrete findings:
  • Ethical reflections clustered around principles like Proportionality and Do No Harm (22.5%), Fairness and Non-Discrimination (16.7%), and Right to Privacy and Data Protection (15.6%), while others like Transparency (3.59%) and Accountability (1.2%) were less salient.
  • Sentiment analysis revealed more concerns (341 comments) than hopes (130), with measured caution emerging as a dominant stance.
  • Pre/post-survey ratings showed a significant decline in scenario favorability (mean change: −0.464), indicating recalibration of attitudes after reflection.
• Advantage over baselines: Demonstrates children's capacity for nuanced ethical reasoning across multiple principles, extending prior work focused on isolated issues. Highlights design fiction as an effective tool for fostering critical reflection and measured caution.
• Experiments / evaluation:
  • Participants: 66 children (aged 10-11) from a UK school.
  • Method: Balanced Incomplete Block Design for scenario exposure; Likert scale surveys and qualitative coding of group discussions.
  • Metrics: Sentiment distribution, pre/post attitudinal changes, thematic analysis of ethical reasoning.
• Limitations and future work:
  • Limited sample size and cultural context may affect generalizability.
  • Abstract principles like transparency and accountability require stronger scaffolding.
  • Future research should test longitudinal effects, cross-cultural applicability, and alternative pedagogical formats.

Summary

This study explores how children perceive and articulate ethical issues related to AI, using speculative design fiction scenarios based on UNESCO principles. Results show uneven distribution of ethical attention, with clustering around principles like fairness and privacy, and thinning in abstract areas like transparency. Children demonstrated measured caution rather than rejection, recalibrating their attitudes after structured reflection. The findings support three recommendations for AI ethics literacy education: addressing multifaceted ethical principles, prioritizing scaffolding for latent areas, and fostering measured caution as ethical growth. This work provides actionable insights for advancing holistic AI ethics literacy education and highlights the potential of speculative methods in fostering critical engagement among children.