Practitioner beliefs and behaviors in AI-enhanced education: DOT framework survey evidence

Synthesis: A 2026 cross-sectional survey (n=72) by Gibson, Azukas, and Knezek examined how higher education practitioners think about and use AI in teaching, grounded in the DOT Framework — a synthesis of Design Thinking and open-systems-theory.

Core Findings

Three-Factor Belief Structure

Exploratory factor analysis of 19 Likert-scale items yielded a three-factor structure with strong reliability (overall α = .90):

1. AI Functional Capabilities (α = .910) — perceived utility of AI tools for teaching tasks
2. Oversight and Governance (α = .79) — need for human supervision, policy frameworks, and critical evaluation
3. Instructor Collaboration and Planning (α = .851) — value of teamwork in AI integration design

This factor structure provides a psychometric anchor for understanding how educators conceptualize AI's role — not as a monolithic "good or bad" judgment but as differentiated beliefs spanning capability, governance, and collaborative practice.

Practice vs. Theory Gap

Practitioners reported frequent use of iterative prompting and content generation, but far less engagement with needs assessment and feedback loops — the front and back ends of a full design cycle. This gap between design-oriented theory and current implementation is the study's central diagnostic finding. It echoes broader patterns in AI Literacy where educators adopt AI for productivity but not yet for systematic instructional redesign.

AI as "Fallible Intern"

The study operationalizes AI as a co-intelligent collaborator following four tenets: (1) AI is fallible — always review, (2) AI supports not replaces higher-order thinking, (3) instructors should actively experiment, (4) instructor maintains agency. This framing aligns with ai-co-intelligence and Human AI Collaboration paradigms where the human remains the epistemic authority.

Institutional Barriers

Widespread lack of policy, training, and infrastructure was reported — consistent with findings across institutional-ai-readiness and AI Governance. Without institutional Scaffolding, even motivated practitioners remain in ad-hoc, fragmented adoption patterns.

The DOT Framework

The DOT Framework integrates Design Thinking stages (Empathize → Define → Ideate → Prototype → Test) with Open Systems concepts (Environment, Input, Process, Structure, Output, Feedback) at both classroom (micro) and institutional (macro) levels. Key intersections:

• Empathize × Environment: Ground instructional design in contextual understanding
• Test × Feedback: Enable recursive improvement at both levels

This study provides the first empirical evidence supporting DOT as a descriptive model — practitioners' beliefs and behaviors partially map to its structure, but the gaps (needs assessment, feedback) reveal where the model is aspirational rather than descriptive of current practice.

What this means for practice

• Instructors. Extend AI use beyond prompting and content generation into both ends of the design cycle: structured prompting with progressive refinement was the most widely used technique (n = 59, 81.9%), while needs assessment and feedback loops lagged.
• Instructors. Treat AI as a fallible intern: review every output, keep the instructor as the epistemic authority, and experiment deliberately rather than delegating higher-order thinking.
• Faculty developers. Prioritize training on evaluating bias and the limits of AI responses: 62.5% (n = 45) ranked it a top priority, against four respondents (5.6%) who called it unimportant.
• Administrators. Treat policy, training, and infrastructure as prerequisites for AI integration rather than add-ons: each institutional barrier was selected by at least 38% of respondents, and no single challenge dominated in isolation.
• Researchers. Start from the 19-item belief scale (overall α = .90) for measurement, but run confirmatory factor analysis before treating the three-factor structure as settled.

Limitations

• The sample (n = 72) is small and self-selected, skewed toward highly engaged AI users in higher-education instructional roles; K-12 educators and student support professionals are underrepresented.
• Self-report data allow social desirability bias, particularly on the items about critical evaluation and oversight of AI outputs.
• The cross-sectional design cannot capture the iterative design, prototype, test, and revise processes the DOT Framework treats as central.
• The three-factor solution is exploratory and hypothesis-generating, ceiling effects in the Oversight and Governance items reduced discriminative power, and some items did not load cleanly onto any factor.

Citation

Gibson, D., Azukas, M. E., & Knezek, G. (2026). Practitioner beliefs and behaviors in AI-enhanced education: DOT framework survey evidence.