HAA Teacher Portal
Human Agency Anchoring · Teacher AI Literacy
HAA — Human Agency Anchoring
Extension of cognition is permitted. Displacement of agency is not.
You've been adapting lessons for your students for years. You know which kid needs extra scaffolding, which explanation fell flat last Tuesday, and which misconception keeps coming back no matter how many times you think you've fixed it. That judgment — the quiet, specific, classroom-trained kind — is exactly what this portal is built to recognise and sharpen. HAA stands for Human Agency Anchoring: in education, AI extends what you can think about and act on, but the agency that makes any of it purposive — goal-setting, quality evaluation, responsibility — stays anchored in you. An “education” in which the learner has not exercised agency is a contradiction in terms, and the same, quieter consequence applies to teaching: a teacher who has delegated pedagogical judgment to an AI has ceded the centre of the practice.
What you bring
- Knowing your class by name
- Reading the room mid-lesson
- Spotting the misconception behind the wrong answer
- Making the call in the moment
- Looking out for your students
What AI can draft
- Lesson outlines
- Worked examples
- Practice questions
- Differentiated versions
- First-pass feedback
Eight days. An honest look at your practice.
Across eight days the portal walks you through a realistic project. Days 1–5 cover the individual teacher's relationship to AI: mapping how you already use it, designing and trying something new with your class, teaching with it, reflecting, and auditing the ethics of your own decisions. Days 6–8 extend the work into the places individual judgment alone cannot go: auditing platform-level features in professional community (Day 6), attending to which students your AI practice disadvantages (Day 7), and rehabilitating assessment validity under AI (Day 8).
It isn't a course. There are no videos to sit through. It isn't a test. Nobody is grading your teaching. It's a structured way to see your own expertise more clearly — and to get sharper at using AI without losing your professional voice, yours or your students'.
What you'll walk away with
- A clear picture of where AI genuinely helps in your subject — and where it quietly makes things worse.
- Prompts and routines you've tested with your actual class, not generic templates from a blog post.
- A language for talking about your AI use with colleagues, leadership, and parents who ask hard questions.
- A considered evaluation across the five areas of teacher AI literacy — evidence you can show, point to, and build on.
The axiom in plain language
Everything in this portal is derived from a single two-clause commitment drawn from the position paper Human Agency Anchoring: An Axiomatic Foundation for AI Competencies in Education (April 2026):
Clause I — Extension.
AI systems are EXTENSIONS of human cognition. They can genuinely participate in thinking by supplying representations, computations, and inferences reliably coupled to you. Cognition can be redistributed across you and the tool — that is a legitimate, long-studied thing.
Clause II — Anchoring.
AGENCY — goal-setting, quality evaluation, bearing responsibility — stays anchored in you. What looks like AI "deciding" is inherited human judgment: choices made by designers and data curators, temporally displaced, replayed by the model. There is no second agent in the room.
From these two clauses, four consequences follow directly: AI has no independent agency (ontological); the right question is never "did the AI produce fluent output?" but "did you come out of this with more capacity that persists when the AI is switched off?" (epistemological); responsibility traces to the human who chose to extend themselves this way (ethical); and three teacher functions — goal-setting, quality evaluation, and responsibility-bearing — cannot be delegated to a tool (pedagogical). From those commitments, the v5 paper derives a six-mode typology of how AI can be coupled to your thinking (Types A–F) and an eight-dimension framework with twenty-four sub-indicators, which the portal's project mirrors across eight days.
Six modes of extension — A through F
The same AI can sit in six very different relationships to your thinking, each with its own anchoring demand. The portal treats them as distinct modes and teaches you to recognise which one you're in. Types A–C are the user-initiated couplings (the original three modes). Types D–F (new in v5) extend the typology to generative, delegated, and ambient couplings — the modes where the chain of agency is most easily broken.
Type A — Task-execution.
You specify the task; the AI executes it. Rubric drafts, content re-levelling, translation. The anchoring demand is precise intent-articulation and evaluation against that intent. Risk: intent under-specified, task done, pedagogical goal unclear.
Type B — Analytical.
The AI produces patterns over data; you interpret and act. Learning analytics, knowledge-state diagnostics, early-warning signals. The anchoring demand is bounded trust — reading the algorithm against your situated knowledge of individual students. Risk: letting the algorithm's framing decide what the situation is.
Type C — Co-cognitive.
You and the AI engage in sustained dialogue that shapes your own thinking. Reflection on teaching, dialogic ideation, critique of your own argument. Among the highest-risk modes, because cognitive outsourcing can feel indistinguishable from cognitive collaboration. Watch for fluent confirmation that sounds like insight.
Type D — Generative / world-producing. new in v5
The AI produces a durable artefact — a simulation, a scenario, a whole environment — that your students subsequently inhabit over time. AI-generated cases, branching narratives, auto-built virtual labs. The anchoring demand shifts to authorship time: you must design with forethought about pedagogical commitments that the environment will then carry, long after you have stopped watching. Risk: hidden pedagogy — affordances of the generated world that you no longer control.
Type E — Proxy / delegated. new in v5
The AI acts on your behalf inside a loop you authorised but do not directly supervise. Autonomous tutors running sessions at 2am, AI graders applied at scale, AI answering student DMs. The anchoring demand is policy-level design — making the chain of authorisation legible so it can be audited and defended. Risk: the teacher authorised the policy but not the act; the student received the act but not the policy. Responsibility is easiest to lose here.
Type F — Ambient / environmental. new in v5
The AI isn't invoked by anyone — it's in the infrastructure. LMS content recommenders, silent re-sequencing, notification systems, dashboards that frame student identity. The anchoring demand is detection itself: agency cannot anchor what it does not see. This is the mode where the work becomes collective — teachers, as a professional community, scrutinising what platforms are doing. Risk: invisibility.
What we believe about teaching
You are the anchoring agent.
Not the AI. Not the framework. Not us. The AI can draft; only you can teach. That is the axiom, not a preference.
Good AI use looks a lot like good teaching.
Diagnose, prescribe, monitor, adjust. The same judgment cycle you've always run — just with some different tools on your desk.
Your subject matters. Your class matters.
Generic advice can't help anyone. The portal treats your subject, your students, and your current unit as non-negotiable context. Without them, the AI can only give back what it took in.
Process counts more than product.
A polished AI lesson with no thought behind it isn't literacy. A messy one you examined carefully is. The evaluative question is always: did YOU come out of this with more capacity?
Voices on teaching
We aren't the first to insist that agency — not fluency — is what education is for. Here are some of the people who've said it better, across centuries, continents, and traditions.
Have a question?
Questions teachers ask — and an assistant who will answer follow-ups
Ten common questions about the portal, its assessment philosophy, and what the AI actually does. Every answer comes with an "Ask a follow-up" chat that is grounded in the HAA philosophy document (which is itself a plain-language rendering of the position paper) — so if our canned answer raises a new question for you, you can interrogate it in your own words.
Read the FAQ →
The one thing we ask of you
Be specific. Not "I used AI to help with a worksheet" — but "I used it to generate acid-base practice questions for 11B1's revision lesson, rejected two because they assumed prior knowledge we haven't covered, and adapted the third for my quieter students."
The more grounded your entries are in your actual classroom, the more the portal can give back. You bring your classroom. We bring the structure. The AI brings drafts you'll throw out, edit, or keep — your call, every time.