Co-Constructing AI Boundaries Framework Component - Inputs

Definition in This Study

Inputs refers to what students uploaded, curated, or selected as context for their AI interactions. This component captures how students framed the task and established the knowledge base from which the AI would draw.

Mollick & Mollick (2023) Connection

The Inputs component aligns with M&M's emphasis on:

Providing context to the AI
Setting clear goals for the interaction
Establishing the parameters of the task

Key M&M Principle: "Provide context and goals to the AI."

The quality and nature of inputs heavily influence the AI's utility, particularly in roles like AI Mentor or AI Tutor, where students must articulate their learning goals and current understanding.

What This Component Analyzes

Primary Focus

Source selection: Which texts, articles, or materials did students upload to NotebookLM?
Curation decisions: What was included vs. excluded?
Problem framing: How did students set up the inquiry task?

Secondary Focus

Evidence of critical source selection (diverse perspectives, challenging materials)
Alignment between inputs and stated learning goals
Constraints or scaffolds built into the input structure

Agency in Inputs: The Knowledge Base Decision

This component captures Agency over Knowledge Base:

Evidence of High Agency	Evidence of Low Agency
Selects diverse, critical sources	Uses only assigned readings
Includes materials that challenge dominant narratives	Limits to safe, mainstream sources
Curates sources strategically to shape AI responses	Uploads everything without curation
Frames problem in complex, nuanced ways	Presents problem simplistically

Boundary-work in Inputs

Students engage in Boundary-work at the input stage by:

Setting epistemological boundaries: What sources count as legitimate?
Limiting AI's knowledge domain: Only providing specific materials
Framing constraints: Directing AI toward particular interpretive lenses

Key Analytic Questions

When coding Inputs, ask:

Curation:
- What sources did they upload?
- Are sources diverse in perspective, method, or ideology?
- Do sources include critical or counter-hegemonic perspectives?
Framing:
- How did they introduce the task to the AI?
- What goals or constraints did they establish?
- Does framing reveal assumptions about knowledge?
Strategic Selection:
- Is there evidence of intentional source selection?
- Do inputs reflect critical engagement with the topic?
- Are inputs designed to test or challenge the AI?

Examples from Data

High Agency in Inputs

Student uploaded:
- 3 course readings (assigned)
- 2 critical articles from scholars of color
- 1 counter-narrative from indigenous perspective
- Personal teaching reflection

Framing to AI: "Using these sources, analyze how literacy
instruction either perpetuates or disrupts deficit thinking
about multilingual learners."

Analysis: Strategic curation bringing critical perspectives; complex framing

Low Agency in Inputs

Student uploaded:
- Course readings only (all assigned)

No explicit framing provided

Analysis: Minimal curation; relies on AI to determine direction

Connection to Epistemic Stance

Input decisions reveal epistemic stance:

Self-authoritative stance: Carefully curates sources to guide AI
AI-authoritative stance: Uploads broadly, trusts AI to sort
Co-constructed stance: Provides diverse sources for collaborative sense-making

Coding Categories for Inputs

Code	Definition	Example
Assigned Only	Only course-assigned materials	Uploads 3 required readings
Strategic Expansion	Adds sources beyond requirements	Includes outside articles
Critical Curation	Deliberately includes counter-narratives	Adds critical race theory text
Minimal Framing	Little/no task setup	Just uploads files
Complex Framing	Detailed problem statement	Multi-part research question

Relationship to Other Framework Components

→ Co-Constructing AI Boundaries Framework Component - Prompts: Inputs establish context; prompts direct interaction
→ Co-Constructing AI Boundaries Framework Component - Outputs: Quality of inputs influences output quality
→ Co-Constructing AI Boundaries Framework Component - Integration: Source selection affects what students integrate

Pedagogical Implications

Supporting strong Input practices:

Teach strategic source selection
Model critical curation (diverse perspectives)
Practice framing complex questions
Discuss how inputs shape AI responses
Reflect on inclusion/exclusion decisions

Data Collection Notes

Where to find evidence:

NotebookLM upload logs (what was uploaded)
Student reflections on source selection
Initial prompt/task framing to AI
Project proposals or planning documents