Student Project Tracks
Big Picture
We are building a system to study how students learn with AI.
The key idea:
Instead of just giving answers, AI should guide how students think—and we should be able to measure that process.
Each project below is designed to lead to publishable research.
Project 1: Epistemic Trace System (Data + Analysis)
What you will build
- A system that records everything students do during an activity:
- answers
- revisions
- AI feedback
- timing
- A structured “event log” of learning:
- who did what
- when
- why (context)
- Tools to:
- export data
- replay sessions
- analyze patterns
Why this matters
This is the foundation of all research.
Without good data → no publishable results
With good data → multiple papers
Paper topics possible:
Paper Direction 1 (Strong, Clean, Publishable)
“Adaptive AI-Guided Collaborative Learning: Analyzing Student Reasoning Through Epistemic Traces”
What this paper does:
- Introduces the system (briefly, but clearly)
- Defines epistemic trace (as part of system, not the focus)
- Shows:
- how guidance works
- how traces are collected
- Then analyzes:
- how students improve over time
- how AI guidance influences that
This is your anchor paper.
Paper Direction 2 (Intervention-Focused)
“The Impact of Timing and Specificity in AI Guidance on Student Problem Solving”
What this paper does:
- Uses your system as the experimental platform
- Varies:
- timing of intervention
- specificity of guidance
- Uses epistemic trace to measure:
- improvement
- time-to-solution
- revision depth
This is where the trace becomes evidence, not the topic.
Paper Direction 3 (Collaboration-Focused)
“Role-Based AI Guidance in Collaborative Learning: Effects on Participation and Reasoning”
What this paper does:
- Uses your role structure
- Applies different guidance per role
- Measures:
- contribution patterns
- reasoning quality
- group dynamics
Again:
- system enables
- trace measures
- paper analyzes
Paper Direction 4 (Process / Learning Trajectory)
“From Attempt to Understanding: Modeling Student Learning Trajectories in AI-Guided Activities”
What this paper does:
- Uses trace to model:
- sequences of attempts
- transitions (wrong → partial → correct)
- Connects that to:
- AI interventions
- timing
This is more AIED/EDM-flavored.
Research questions
- How do students improve their answers over time?
- What does “getting stuck” look like?
- What patterns indicate real learning?
Best fit
- You like systems, structure, and data
- You want to do serious research
Project 2: Adaptive AI Guidance System
What you will build
- A system where AI gives help based on:
- time spent
- student behavior
- previous attempts
- Guidance that evolves:
- early: open-ended
- middle: hints
- late: more direct help
- A configurable system (YAML-based) for:
- different guidance strategies
- role-based prompts (facilitator, recorder, etc.)
Why this matters
This is how we test:
Does AI actually help students learn better?
Research questions
- When should AI intervene?
- Does timing matter?
- Does more detailed guidance help or hurt?
Possible paper topics
- “Adaptive AI Guidance in Collaborative Learning: Effects of Timing and Feedback Specificity”
- “Designing Time-Based AI Interventions to Support Student Problem Solving”
Intervention / Experiment Papers
- “When Should AI Help? Evaluating the Timing of Interventions in AI-Supported Learning”
- “From Hints to Guidance: The Impact of Progressive AI Scaffolding on Student Performance”
Role-Based / Structured Guidance
- “Role-Based AI Guidance in Collaborative Learning Environments”
- “Supporting Group Roles with AI: Effects on Participation and Learning Outcomes”
More Ambitious / AIED-Level Direction
- “Adaptive Scaffolding with AI: Modeling and Evaluating Dynamic Guidance Strategies in Collaborative Learning”
- “Balancing Autonomy and Support: An Experimental Study of AI Intervention Strategies in Student Learning”
Best fit
- You like AI behavior and experimentation
- You want to study how people learn
Important Notes
- These projects are deeply connected
- Everything must be logged for research
- Each project should lead to a paper
What success looks like
By the end of the summer:
- A working system used in real activities
- Data collected from student interactions
- Initial analysis results
- Draft of a research paper
How to choose
Pick the one that fits how you think:
- Want to understand what students are doing → Project 1
- Want to understand how AI should guide students → Project 2