Case ReportPublished empirical studyJun. 23, 2025
AAB-CASE-2025-RV-037
Children's Mental Models of AI Reasoning: Implications for AI Literacy Education
IDC 2025 UW study on children AI reasoning models.
This page documents an AI literacy or AI education case for registry purposes. It is descriptive and does not imply AAB endorsement of any specific tool, provider, or intervention.
01
Implementation
University of Washington
02
Learning context
In-school (K–12)
03
AI role
Evaluator
04
Outcome signal
Mental models
Registry Facets
0
Education Level
- K-5
- 6-8
Subject Area
- AI literacy
Use Case Type
- Field study
- Co-design
Stakeholder Group
- Students
AI Capability Type
- Reasoning
Implementation Model
- Classroom-level
Evidence Type
- Mixed methods
Outcomes Domain
- Mental models
Implementing Organization
1
Organization Type
University of Washington
Location
Washington, USA
Primary Facilitator Role
Research team
Learning Context
2
Setting Type
- In-school (K–12)
Session Format
Co-design + field sessions
Duration
Two phases
Group Size
8 + 106
Devices
ARC-like puzzles
Constraints
- Specific puzzle paradigm
- Regional sample
Learner Profile
3
Age Range
Grades 3–8
Prior AI Exposure Assumed
Chatbot and LRM awareness
Prior Programming Background Assumed
Not required
Educational Intent
4
Primary Learning Goals
- Identify mental models of AI reasoning
- Compare ages
- Scaffold AI4K12 reasoning
Secondary Learning Goals
- Inform XAI for youth
What This Was Not
- Not ARC engineering course
AI Tool Description
5
Tool Type
ARC-inspired reasoning tasks
AI Role
- Evaluator
Languages
English
User Interaction Model
- Elicit explanations of AI solution behavior
Safeguards
- Avoid sentience myths
- Child ethics
Activity Design
6
Activity Flow
- Co-design
- Field protocol
- Coding models
Human Vs AI Responsibilities
- Educators teach limits of statistical AI
Scaffolding Strategies
- Structured puzzles externalize reasoning
Observed Challenges
7
Educators Reported
- Developmental differences in inherent vs pattern views
- Tensions in teaching reasoning with LRM hype
Design Adaptations
8
Adaptations
- Leverages ARC culture for literacy not only benchmarks
Reported Outcomes
9
Engagement
- Broad grade span
Learning Signals
- Clear developmental patterns
Educators Reflection
Curriculum and XAI design implications.
Ethical & Privacy Considerations
10
Privacy
- Secure child data
- Non-stigmatizing framing
Evidence Type
11
Evidence
- Post assessment
- Activity documentation
- Practitioner observation
Relevance to Research
12
Potential Research Use
- Validated mental-model instruments
- Link to learning gains
Relevant Research Domains
- AI reasoning education
- Developmental psychology
Case Status
13
Case Status
- Completed
AAB Classification Tags
14
Age
3–8 grades
Setting
US
AI Function
Reasoning literacy
Pedagogy
Participatory
Risk Level
Low
Data Sensitivity
Medium
Registry Metadata
15
Case ID
AAB-CASE-2025-RV-037
Publication Status
Published empirical study
Tags
caseK-5Washington, USAClassroom-levelReasoningAI literacyField studyCo-design