Transparent evidence and governance
ExperienceCraftAI makes its pedagogical checks, design evidence and human decision points visible. This turns AI generation from a black box into an auditable co-design process.
Alignment evidence matrix
Every claim is connected to a visible activity design feature
| Framework / criterion | Target | Design evidence | Confidence |
|---|---|---|---|
| Learning outcome | Classify microorganism groups using characteristics | Evidence-based classification and defence | High · 96 |
| Bloom: Apply | Use characteristics in a novel scenario | Outbreak clue analysis | High · 95 |
| Bloom: Analyse | Differentiate plausible organisms | Compare evidence and reject an alternative | High · 97 |
| Bloom: Evaluate | Justify a response plan | Team presentation and cross-questioning | High · 91 |
| 7 Modes: Doing | Learn through purposeful action | Hands-on sorting and investigation | High · 94 |
| 7 Modes: Social belonging | Construct understanding with others | Interdependent team roles | High · 93 |
| Seriously Fun: Agency | Meaningful learner choices | Teams choose evidence strategy and response plan | Good · 87 |
| Feasibility | Deliverable in 120 minutes | Timed run sheet and low-complexity materials | High · 96 |
Human oversight trail
Clear accountability
AI is advisory, not autonomous
The system proposes, explains and checks. The lecturer owns the teaching decision and can edit every output.
Traceable design rationale
Why each element exists
The outbreak scenario directly answers the prior feedback that learning felt too theoretical. Team roles provide structured social interaction, while the response plan raises the task from classification to evaluation.
Constraint assurance
Before generation and after refinement
No live microorganisms are used. Materials are printable, instructions are multimodal and no success cue depends on colour alone.
Evidence-informed improvement
Post-implementation learning
Student outputs, rapid sentiment, timing data and lecturer notes are combined to propose specific, reviewable refinements for the next run.