The Barker Institute & University of Sydney: Young People, Learning, and Generative AI

Author

Natasha Arthars, Hongzhi (Veronica) Yang, Matthew Hill, Danny Liu, & Lina Markauskaite

Young People, Learning, and Generative AI: PreK–12 Literature Review

Primary Source: The Barker Institute Report Page
Webinar: Young People, Learning, and Generative AI (YouTube)

Overview

This rapid literature review from The Barker Institute and the University of Sydney synthesises emerging evidence on generative AI (GenAI), young people, learning, and development in PreK–12 education. The study screened 975 publications and analyzed 271 studies.

The review adopts an ecological framework: GenAI must be understood as part of the broader environments in which learners think, participate, and judge, rather than as an isolated technology.

Core Findings

  • Affective Gains vs. Durable Learning: GenAI frequently increases student interest, motivation, and confidence while reducing learning anxiety. However, these affective gains do not automatically translate to durable, retained learning once AI scaffolding is removed.
  • Task Completion vs. Skill Retention: GenAI shows strongest evidence for boosting immediate task performance and product quality. Unstructured access, however, risks answer-seeking, dependence, and judgment offloading.
  • The 5 Ecological Conditions: Learning outcomes depend on the interaction among five factors:
    1. Learners: Prior knowledge, proficiency, self-regulation, trust, and developmental stage.
    2. GenAI Tool Design: Answer-oriented versus scaffold-oriented interactions, domain specificity, and interface cues.
    3. Learning Task: Pedagogical role, expected product, and distribution of responsibility.
    4. Social Arrangements: Teacher-peer-AI feedback and participation dynamics.
    5. Cultural & Institutional Context: Access, norms, authority, and expectations.
  • Four Pedagogical Functions: Recommends structuring AI integration around whether students learn from AI, with AI, about AI, or by shaping AI tools.

Primary Source References