It differs from generic AI use in education because it is anchored to an explicit instructional model — Bloom’s taxonomy, constructive alignment, inquiry, competency-based progression — and to a defined curriculum standard, rather than producing plausible content with no pedagogical structure behind it. Edves operationalises 14+ research-driven instructional models this way.
The problem it addresses
Generative AI arrived in schools before any settled view of how it should be used. Two failure modes followed quickly. In the first, teachers use a general-purpose chatbot to produce lesson material that is fluent but pedagogically arbitrary — it has no cognitive progression, no alignment to the standard being taught, and no relationship to what this particular class already knows. In the second, schools ban the tools outright and lose the productivity gain entirely.
Pedagogy with AI is the middle position, and it is a deliberately narrow one: AI generates, the teacher decides.
Definition
Pedagogy with AI is the practice of using artificial intelligence to design, adapt and evidence instruction while the teacher retains professional judgement and accountability. Three conditions distinguish it from unstructured AI use:
- Model-anchored. Every generated artefact is produced against an explicit instructional model — Bloom’s taxonomy, Depth of Knowledge, Understanding by Design, SOLO, the Kolb cycle — rather than to a generic prompt.
- Standard-anchored. Output is mapped to a named curriculum standard, so a lesson can be traced to the specific objective it serves and coverage gaps become visible.
- Teacher-owned. The teacher edits, approves and signs off. The system records that they did. Accountability does not transfer to the model.
How it differs from adjacent ideas
| Approach | What the AI does | Who is accountable |
|---|---|---|
| AI tutoring | Interacts directly with the student | Ambiguous; the model mediates learning |
| Adaptive learning | Sequences content by prior performance | The algorithm sets the path |
| Generic AI use | Produces material on request | Teacher, but with no structure to check against |
| Pedagogy with AI | Drafts instruction against a stated model and standard | The teacher, explicitly and on the record |
What it looks like in practice
1. Lesson design
The teacher selects a curriculum objective, a class, and an instructional model. The system drafts a lesson sequence with the cognitive level of each task tagged, differentiation for the range of prior attainment in that class, and the assessment that will evidence the objective. The teacher edits it. What ships is the teacher’s lesson.
2. Real-time adaptation
During teaching, responses to checkpoint questions indicate whether the class has met the objective. The system proposes an adjustment — reteach with a different representation, extend, or move on — and the teacher chooses.
3. Observation and professional growth
Teaching is observed against a rubric in which every rating requires evidence from the room. Vague, hedged feedback is stripped out. The observation record is defensible and the teacher has a right of reply. See higher education for the university implementation of the same engine.
4. Assessment
Items are generated against the standard and the cognitive level, not just the topic. Results feed back into what the next lesson needs to address. See assessment.
The 14+ instructional models
Edves operationalises a defined set of models rather than treating pedagogy as a free-text field. Each carries its own prompt structures, task shapes and rubric logic.
| Model | What it structures |
|---|---|
| Experiential learning | Hands-on cycles with structured reflection |
| Project-based learning | Scaffolded milestones and rubric feedback |
| Bloom’s taxonomy | Cognitive level tagging across tasks |
| Competency-based learning | Mastery tracking and personalised progression |
| Gamified learning | Engagement mechanics tied to mastery, not activity |
| Problem-based learning | Scenario-driven scaffolds |
| Inquiry-based learning | Questioning sequences and investigation design |
| Design thinking | Iterative prototyping and peer critique |
| Spaced learning | Retrieval scheduling for retention |
| Micro learning | Short bursts with immediate feedback |
| Blended learning | Synchronous and asynchronous rotation |
| SEND pedagogies | Scaffolds for ASD, dyslexia and ADHD |
| Assessment for learning | Formative cycles with next-step planning |
| Career pathways | CTE and T-Level aligned progression |
Why the anchoring matters
An unanchored model will produce a lesson plan that reads well and teaches badly — six activities at the same cognitive level, no retrieval, no check for understanding, no relationship to the standard. The output is fluent, which is exactly what makes it hard to spot. Anchoring to a model and a standard turns a plausibility engine into something a head of department can audit.
It also produces the record. When every lesson is traceable to an objective and every observation to evidence, a school can answer inspection and accreditation questions from data it already holds rather than from a scramble.
Implementation
Schools typically adopt in three stages: lesson design first, because the time saving is immediate and visible; then observation and CPD, which requires trust and a clear agreement that observation is developmental rather than punitive; then assessment, which touches the most stakeholders. Attempting all three at once is the most common cause of stalled adoption.
Regional guidance: United States, United Kingdom, Nigeria, Ghana.