HAA is Human Agency Anchoring. Earlier drafts used "Human Agency Augmentation" and that phrasing may still appear in a few corners of the portal; the current position paper settles on "Anchoring" because the word specifies both what is anchored (agency, in Bandura's strict sense — goal-setting, forethought, self-reactiveness, self-reflectiveness) and how (as the purposive anchor of the extended activity). The authoritative statement is the v5 position paper by Xiangen Hu (The Hong Kong Polytechnic University), available at /HAA_Axiom.
The axiom has two clauses. Clause I: in educational contexts, AI systems function as extensions of human cognition — they can genuinely participate in thinking by supplying representations, computations, and inferences reliably coupled to you. Clause II: the agency that makes any such extension purposive — setting goals, evaluating outcomes, bearing responsibility — stays anchored in human agents. What AI exhibits as "deciding" is inherited human judgment replayed from training, not a second agent in the room.
From the axiom, four corollaries, a six-mode typology of extension (Types A–F), and an eight-dimension competency framework with twenty-four sub-indicators follow. Types A–C are user-initiated couplings (task-execution, analytical, co-cognitive); Types D–F cover generative/world-producing, proxy/delegated, and ambient/environmental couplings where the chain of agency is most easily broken. Dimensions I–V cover the individual teacher's practice; Dimensions VI–VIII cover collective agency, equity-centred practice, and assessment redesign under AI.
Every design choice in this portal — the eight-day project, the Domain Floor that caps non-subject-specific answers at Level 2, the coach-vs-editor split, the effort-tracking table, the quality monitor strip — is a concrete consequence of that axiom. Extension of cognition is permitted. Displacement of agency is not.
No — the opposite. The whole portal is built on the conviction that AI cannot teach your class. It can draft worksheets, propose questions, and summarise texts. It cannot read the room, notice which student is struggling today, or decide what tomorrow's lesson should cover. The five days you spend here are about making that boundary obvious — to you, and to anyone who asks.
You're being recognised, not graded. The rubric looks for evidence of professional judgment across five areas, and your results belong to you. There's no pass, no fail, no public ranking, and no comparison against other teachers. The only person who sees your scores is you — and, if you're part of a cohort administered by your institution, whoever your institution has specifically authorised.
No. If you've typed a prompt into ChatGPT and thought "that wasn't quite right" — you're more ready for this than someone who uses AI daily but never questions what comes back. The portal is built for your kind of thinking, not for tech enthusiasm.
That's a legitimate professional position, and the portal has a place for it. Several of the log types — especially the ethical audit — exist to record exactly this kind of reasoned refusal.
A teacher who examined a tool and said "not for my Year 9 English class, here's why" is demonstrating more AI literacy than one who adopted it without thinking. The rubric credits principled non-use, not just use.
Because advice that doesn't know your context isn't advice — it's noise. An AI that doesn't know you teach Year 11 Chemistry can only produce generic "use AI to engage students" nonsense, and you already know how useless that is.
The profile wizard exists so the system can give you something specific enough to be worth arguing with.
Around 30–45 minutes per day for five days. Most of that time is you doing things you would be doing anyway — planning, teaching, reflecting — and keeping short notes as you go. This is not a second job on top of your real one.
Your daily logs and coaching conversations are yours. If your institution set up an admin account to manage a cohort, admins see aggregate data across the group; they can only drill into an individual if that individual has been explicitly flagged for support.
Your pedagogical reasoning is not performance-review material, and the system is built that way on purpose.
Three things, all in service of your judgment:
It never makes decisions for you, grades you silently, or hands your entries to the model provider for training.
Say so in your log. Reasoned disagreement is one of the strongest pieces of evidence the rubric looks for. A teacher who read the coach's suggestion, rejected it with a specific pedagogical reason, and wrote that down is demonstrating more literacy than one who accepted it passively. The AI is a colleague with drafts, not an authority with answers.