Kevin Brooks completed his five-day HAA project with genuine enthusiasm for AI tools but consistently shallow engagement with the framework's deeper demands. His work illustrates the Level-2 profile: someone who uses the right vocabulary, follows instructions, and produces complete submissions — but whose understanding of AI in teaching remains tool-centric rather than pedagogically grounded.
On Day 1, Kevin's Extension System Analysis classified five activities into A/B/C types, but the classifications mapped to tool names rather than cognitive functions. 'ChatGPT for brainstorming' was labeled Type A and 'Grammarly for editing' was labeled Type A, but neither classification explained what cognitive function was being extended or how the teacher's pedagogical role changed. The 'If AI were removed?' responses consistently defaulted to 'I would do it myself, just slower' — a response that is technically true for A-type but reveals no analysis of whether the activity's pedagogical function would change. His Boundary Map listed real limits ('AI sometimes gives wrong answers about plot details') but at a level of generality that could apply to any subject — none of the five entries named a specific lesson activity, a specific learning objective, or a specific student population. The Teacher Irreplaceability Declaration was where the gap was most visible: all five reasons ('building relationships,' 'classroom management,' 'motivating students,' 'understanding emotional needs,' 'making judgment calls about grades') are true of all teachers in all subjects. Not one reason referenced English as a discipline, literary analysis as a cognitive activity, or the specific interpretive challenges of teaching Brave New World. The Mental Model Audit was the strongest Phase 1 piece — Kevin identified genuine overstatements in vendor copy — but even here, his analysis stayed at the level of 'that's a stretch' rather than identifying the structural assumptions about teacher agency embedded in the marketing language.
Day 2's Prompt Engineering Log showed genuine iterative improvement from 'give me discussion questions' to a structured prompt specifying chapter range, difficulty levels, and question types. However, the revision rationale at each stage noted surface-level adjustments ('need harder questions,' 'should mention specific chapters,' 'ask for different types') rather than pedagogical functions that were or were not being served. No iteration explained what the questions needed to achieve in terms of student learning about Huxley's themes or literary analysis skills. The Output Evaluation Reports were Kevin's strongest work: he caught a genuine hallucination (misplaced plot event), identified a one-dimensional character analysis, and flagged an overconfident thematic claim. But even these evaluations stopped at 'students would get wrong information' rather than naming the specific misconception a Grade 9 reader would develop or the specific interpretive skill that would be undermined. The Customised Artefact (Discussion Prompt Generator) was a functional template but essentially a format specification — it did not embed any theory about what makes a discussion question pedagogically productive for this unit.
Day 3 was Kevin's most promising phase. His micro-lesson included a genuinely good moment — having students find errors in an AI-generated summary — and his decision log showed real-time adjustments (redirecting a student from asking 'what's the answer' to asking 'why'). The Student Guidance episode touched on important points ('AI gives you its interpretation, not THE interpretation') but never specified what kinds of errors students should watch for in literary analysis specifically. Kevin's guidance could apply equally well to any subject. The student work evaluations showed Kevin can identify when students are absorbing AI framing uncritically, but his feedback responses ('try rewriting in your own words') addressed the symptom (AI voice in student writing) rather than the cause (students lack their own interpretive framework for the novel).
Day 4 revealed the most telling pattern: Kevin accepted all five AI-generated insights about his practice, modifying only one and rejecting none. The insights were generic praise ('you show strong critical thinking,' 'your prompt engineering shows progressive refinement,' 'your ethical awareness is developing') and Kevin accepted them at face value without noting the echo-chamber dynamic — the AI was reflecting his own framework language back to him, not independently validating his practice. His three stakeholder communications were appropriate in tone but used nearly identical content across all three audiences, differing mainly in formality level rather than in what was communicated or why. His PD plan goals were all AI-proposed or self-identified at the tool level ('learn more prompting techniques,' 'explore AI tools for feedback') rather than at the pedagogical level.
Day 5's ethical audit listed real categories of risk (dependency, narrowing of interpretations, academic integrity, privacy, platform lock-in, cultural bias) but each risk was stated at the level of general principle rather than grounded in a specific lesson activity, a specific student population, or a causal mechanism. 'Students might become dependent on AI for brainstorming' is a valid concern but it does not name which students, in which activity, through what mechanism, with what specific consequence for their development as readers and writers of literary analysis. The Purpose Drift Audit was honest — Kevin acknowledged that vocabulary generation and quiz creation were efficiency-driven — but his revisions were modest ('will go back to selecting vocabulary myself,' 'write my own key questions'). No revision changed a tool or platform; all three unit plan changes were behavioral adjustments within existing tools.
Kevin's project demonstrates the gap between AI fluency and AI literacy. He can use the tools competently, follows the project structure fully, and produces work that meets minimum requirements at every stage. What is consistently missing is the subject-specific pedagogical reasoning that would make his analysis distinctively that of an English teacher teaching Brave New World — rather than any teacher using AI tools in any class.
100 log entries across 6 types
Click any task to view the full draft in the workspace wizard