How does agentic coding develop?
We examine how learners plan, prompt, inspect, test, revise, and reflect across projects and over time.
Research agenda
Studying learning when coding includes an agent.
We investigate how youth develop computational agency while planning, prompting, testing, revising, and reflecting on creative computational work. The questions on this page describe an active research agenda—not findings.
This work builds on the UCI Digital Learning Lab’s broader Computing and AI for All research into meaningful, creative participation in emerging forms of computing and artificial intelligence.
Research focus
Learning process, learner development, and learning context.
The project connects close attention to what young people do with questions about what they learn and how environments support that learning.
What does ownership look like?
We ask how learners express intent, reason about generated work, make creative choices, and remain responsible for results.
What supports participation?
We study the roles of educators, peers, community, professional learning, activities, and platform design.
Areas of inquiry
Questions that guide design and study.
- How do prompting, testing, revision, and reflection connect over time?
- How can a coding agent support learning without displacing learner judgment?
- How do youth develop ownership of AI-supported computational artifacts?
- How does educator facilitation shape productive work with coding agents?
- How do experiences differ across learners, projects, and informal settings?
Learning framework
Ideas for understanding creation, participation, and guidance.
The framework brings established learning perspectives into conversation with the emerging practice of agentic coding.
Constructionism
People learn through making personally meaningful, shareable artifacts.
Situated learning
Knowledge develops through participation in authentic practices and communities.
Cognitive apprenticeship
Modeling, coaching, scaffolding, and reflection make expert practices more visible.
Computational agency
Learners direct computational tools toward meaningful goals and take responsibility for decisions.
Informal STEM settings
Research embedded in the places where youth create.
The work is designed especially for interest-driven, community-based learning beyond conventional school curriculum.
Why context is part of the research
Informal environments vary in their cultures, resources, facilitation, and pathways into participation. Rather than treating setting as background, the project studies how those conditions shape agentic coding.
- Activity design
- Educator facilitation
- Peer collaboration
- Local adaptation
- Access and participation
Design-based implementation research
Design, study, learn, and refine.
The project uses iterative cycles of co-design, implementation, data collection, analysis, and refinement in collaboration with learning organizations.
This approach connects platform and activity development with real implementation conditions. Emerging evidence can inform the next design cycle while collaborators contribute knowledge of their communities and programs.
Data & evidence
What the project plans to study.
Potential sources are considered together and used under appropriate research and human-subjects procedures.
Creative artifacts
Youth projects and how their features develop through revision.
Interaction traces
Prompts, agent interactions, testing, and revision patterns captured by the platform.
Learning in context
Observations, interviews, activities, educator perspectives, and implementation records.
Three-year trajectory
Learning locally, refining, and expanding.
The grant outlines a staged direction that can respond to what the project learns. Later phases remain intentionally flexible.
- Year 1
Initial implementation
Begin learning and research with two partner sites.
2 initial sites - Year 2
Expansion
Refine the learning environment and research approach based on Year 1, then add more partner learning sites or centers.
More partner sites - Year 3
Broader national reach
Work toward a broader network of learning partners and the possibility of national-scale implementation.
Broader national reach