Co-Constructing AI Boundaries Framework Component - Prompts

Definition in This Study

Prompts refers to how students directed, constrained, or collaborated with the AI model through their questions, commands, and conversational moves. This component captures the specific language and strategies students used to shape the AI's responses.

Mollick & Mollick (2023) Connection

The Prompts component mirrors M&M's core mechanism of interaction:

Assignment roles (Tutor, Coach, Mentor, Teammate, etc.)
The function of prompting to shape AI behavior
Strategic use of constraints and direction

Key M&M Principle: Prompts are the central tool for defining and controlling the AI interaction.

The entire M&M paper is subtitled "with Prompts" and dedicates sections to example prompts for all seven approaches, underscoring the critical role of prompting in shaping learning experiences.

What This Component Analyzes

Primary Focus

Prompt complexity: Simple/generic vs. complex/specific
Constraint level: Open-ended vs. constrained
Cognitive demand: What kind of thinking does the prompt require?
Role assignment: Does student assign the AI a specific role?

Secondary Focus

Evidence of iteration (revising prompts based on outputs)
Metacognitive awareness in prompt design
Strategic use of prompt engineering techniques

Agency in Prompts: The Cognitive Task Decision

This component captures Agency over Cognitive Task:

Evidence of High Agency	Evidence of Low Agency
Complex, multi-step prompts	Generic, single-step prompts
Constraining language (boundaries)	Open-ended, vague requests
Demands critique, synthesis, comparison	Asks for summary or generation
Assigns specific roles to AI	No role assignment
Iterates on prompts strategically	One-shot prompting

Prompt Types and Cognitive Demand

Low Cognitive Demand (Low Agency)

Summarization: "Summarize this article"
Generation: "Write a paragraph about..."
Definition: "What is critical literacy?"

Moderate Cognitive Demand

Comparison: "Compare Freire and hooks on literacy"
Application: "Apply this theory to my teaching context"
Organization: "Organize these themes"

High Cognitive Demand (High Agency)

Critique: "Critique this argument using critical race theory"
Synthesis: "Synthesize these three perspectives into a coherent framework"
Evaluation: "Identify logical fallacies in this text"
Devil's Advocate: "Challenge my thesis from a [specific] perspective"

Boundary-work in Prompts

Students engage in Boundary-work through:

Constraining AI capabilities: "Only use these sources; don't search the web"
Setting behavioral boundaries: "Don't write the paper; help me think"
Limiting AI authority: "Suggest options, but I'll decide"
Demanding transparency: "Explain your reasoning"

M&M Assignment Roles in Student Prompts

M&M Role	Purpose	Example Student Prompt
AI as Tutor	Explain concepts, answer questions	"Explain the zone of proximal development"
AI as Coach	Prompt metacognition, guide process	"What questions should I ask myself as I analyze this?"
AI as Mentor	Long-term guidance, goal-setting	"Help me plan my research project over the next month"
AI as Teammate	Collaborative problem-solving	"Let's brainstorm solutions together"
AI as Student	Student teaches AI to deepen learning	"I'll teach you about critical literacy"
AI as Simulator	Practice scenarios or dialogues	"Simulate a parent-teacher conference"
AI as Tool	Mechanical task completion	"Check this text for grammar errors"

Key Analytic Questions

When coding Prompts, ask:

Complexity:
- Is the prompt simple or multi-layered?
- Does it require multiple cognitive operations?
Constraint:
- Does student limit AI's scope or behavior?
- Are boundaries explicitly set?
Cognitive Demand:
- What level of thinking does this prompt require from the AI?
- Summarize? Critique? Synthesize?
Role Assignment:
- Does student assign AI a specific role?
- Which M&M role (if any)?
Strategic Intent:
- Is there evidence of prompt engineering knowledge?
- Does student revise prompts based on results?

Examples from Data

High Agency Prompt

"Act as a critical reviewer trained in critical race theory.
Analyze this lesson plan and identify where it might
inadvertently perpetuate deficit thinking about students
of color. Provide specific examples and suggest revisions."

Analysis:

High complexity (multiple steps)
Constraining (specific theoretical lens)
High cognitive demand (critique, application)
Clear role assignment (critical reviewer)

Low Agency Prompt

"Summarize these articles."

Analysis:

Low complexity (single step)
No constraints
Low cognitive demand (summarization)
No role assignment

Connection to Epistemic Stance

Prompt design reveals epistemic stance:

Self-authoritative: Uses AI as thinking partner, not answer generator
AI-authoritative: Asks AI to produce final knowledge products
Co-constructed: Engages in dialogue, iterates collaboratively

Coding Categories for Prompts

Code	Definition	Example
Generic Ask	Simple, common prompt	"Summarize this"
Role Assignment	Assigns specific role	"Act as a literacy coach"
Constraining	Sets explicit boundaries	"Only use these sources"
Devil's Advocate	Requests critical challenge	"Challenge my argument"
Synthesis	Requests integration of multiple ideas	"Synthesize these three theories"
Critique	Requests critical analysis	"Critique this from a feminist lens"
Iterative Refinement	Revises prompt based on output	"Actually, focus more on..."

Relationship to Other Framework Components

← Co-Constructing AI Boundaries Framework Component - Inputs: Inputs provide context; prompts direct action
→ Co-Constructing AI Boundaries Framework Component - Outputs: Prompt quality influences output quality
→ Co-Constructing AI Boundaries Framework Component - Integration: High-demand prompts may yield more useful outputs

Pedagogical Implications

Teaching effective prompting:

Model prompt progression (simple → complex)
Teach constraint setting for boundary-work
Practice assigning roles strategically
Discuss cognitive demand in prompts
Reflect on prompt-output relationships
Encourage iteration and refinement

Data Collection Notes

Where to find evidence:

NotebookLM chat logs (every prompt)
Student reflections on prompting strategies
Iteration patterns (revisions to prompts)
Explicit metacommentary about prompting