Implementing Organization
1
Organization Type
Higher Education (Art University)
Organizations
- Faculty of Applied Sciences, Macao Polytechnic University
- Art and Design School, Anhui Broadcasting Movie and Television College
Location
Eastern China (art university setting)
Learning Context
2
Setting Type
- Informal learning
- Afterschool center
- In-class (Higher Ed)
- Private program
Session Format
13-week quasi-experimental study
Duration
Weekly 90-minute sessions over 13 weeks
Devices
Personal mobile devices or lab computers (web interface)
Constraints
- No individual logins required beyond university authentication
- No personal data collected outside study parameters
- Classroom-based setting with instructor supervision
- Fixed class schedule (weekly 90-minute sessions)
Learner Profile (Non-identifiable)
3
Age Range
Approx. 19–23 years (undergraduate students)
Program
Digital Media Art
Prior AI Exposure (Assumed)
No prior experience with generative AI tools assumed
Prior Programming Background
None assumed
Educational Intent
4
Primary Learning Goals
- Mastery of introductory digital media art concepts
- Development of self-regulated learning skills (planning, monitoring, reflection)
Secondary Learning Goals
- Awareness of AI as a learning support tool
- Ability to critically evaluate AI-generated responses
- Improved learning attitudes and technology acceptance
What This Was Not
- Not a programming lesson
- Not an AI theory lesson
- Not an assessment-driven activity
AI Tool Description
5
Tool Name
ArtChat
Tool Type
Curriculum-aligned generative AI chatbot
AI Role
ArtChat functions as an interactive learning assistant that provides course-specific, context-aware responses to student queries, supporting knowledge construction and self-regulated learning.
User Interaction Model
- Students log into ArtChat via web interface
- Students type text-based questions related to course content
- ArtChat retrieves relevant information using RAG and knowledge graphs from uploaded course materials
- Students receive structured, citation-backed answers
- Students may ask follow-up questions or request clarification
Safeguards
- Knowledge base restricted to instructor-uploaded course materials
- No access to external or unfiltered internet content
- Content moderation via AI model alignment
- Instructor oversight during class sessions
Activity Design
6
Activity Flow
- Instructor introduces weekly topic
- Students engage with ArtChat during or after class for self-directed learning
- Students reflect on responses and integrate them into understanding
- Instructor facilitates discussion and addresses common questions
Scaffolding Strategies
- Instructor guidance on effective prompting
- Reflection prompts embedded in study design
- Peer discussion and instructor-led synthesis
Human vs AI Responsibilities
7
Human
Question formulation, critical evaluation, reflection, discussion
AI
Information retrieval, contextual explanation, draft generation
Observed Challenges
8
- Some students initially over-relied on AI without critical evaluation
- Vague or poorly structured queries led to less useful responses
- Time constraints limited deeper exploration in some sessions
- General LLM group occasionally received off-topic or inaccurate information
Design Adaptations Made
9
- Added in-class orientation on effective questioning strategies
- Reinforced "AI as a learning partner, not an answer machine" framing
- Encouraged students to compare AI responses with course materials
- Integrated reflection activities into post-class assignments
Reported Outcomes
10
Engagement
High participation; ArtChat group demonstrated sustained interaction with the tool and increased voluntary engagement with course content
Learning Signals (Qualitative)
- Students revised their understanding based on AI feedback
- Some questioned or challenged AI responses, prompting deeper inquiry
- Peer discussions about AI-generated content increased
- Students reported feeling more in control of their learning process
Facilitator Reflection
"ArtChat didn't just give answers—it gave students a reason to ask better questions. The most meaningful learning happened when they reflected on why the AI responded the way it did."
Ethical & Privacy Considerations
11
- No personal data collected beyond basic demographic information for research purposes
- All student data anonymized in analysis and reporting
- Informed consent obtained from all participants
- Study approved by institutional ethics review
- Tool usage aligned with university IT policies
Relevance to AI Education Research
12
Potential Research Use
- Empirical validation of AI tools in higher education
- Self-regulated learning and AI interaction studies
- Cross-disciplinary AI applications (AI + education + arts)
Relevant Research Domains
- Learning sciences
- Educational technology
- AI literacy
- Art education
- Human-AI collaboration
Contributors
13
Author
Yang, Chenglin; Un, Kin-Seong; Jiang, Shujing; Tan, Tao; Lam, Chi-Kin; Sun, Yue
Editor
Cao, Christopher — Research Assistant, M.S. Student, California State University, Fullerton