In the classroom
Activities and examples
AIPLA combines guided tutors with prepared learning material. The material may be a problem, document, simulation, experimental procedure, graph, data table, image, or concept map. The tutor can refer to what the student is doing rather than treating the conversation as an isolated chat.
The paired tutor and workbench pattern
A central AIPLA pattern is to design the tutor and the interactive activity together. A tutor can ask about a particular graph, slider, piece of equipment, or stage in an experiment because those elements are part of its teaching context.
A typical sequence is:
- Predict what will happen and explain why.
- Explore by changing a parameter, inspecting evidence, or carrying out a procedure.
- Observe the result in a graph, diagram, measurement, simulation, or document.
- Reflect through questions from the tutor and discussion within the group.
- Summarise the physics in the students' own words or representations.
This is not a fixed recipe for every lesson. Some activities need no simulation, while others may emphasise experimental decisions, conceptual dialogue, or critique of an AI-generated explanation.
Three developed case studies
Boldkast: projectile motion
Boldkast pairs a Danish Socratic tutor with a projectile-motion workbench. Students vary launch conditions, compare trajectories and representations, and work through a structured physics problem.
The activity established the platform's core feedback loop: student interaction with the workbench becomes visible context for the tutor, and the tutor can use that context in its next question.
LED Planck: procedural virtual laboratory
LED Planck explores a different form of interaction. Students rehearse and interpret a simplified experiment involving an LED circuit, threshold voltage, wavelength, and an estimate of Planck's constant.
The case study helps distinguish what a virtual laboratory can support from what must remain grounded in real equipment, measurement uncertainty, and classroom laboratory practice.
Read the LED Planck case study.
KineBot: migrating an existing teaching artefact
KineBot brings together kinematics simulations, graphs, conceptual prompts, and structured activities. Its migration into AIPLA established a reusable intake process for externally created teaching artefacts.
The original standalone prototype included its own browser-side AI integration. The maintained AIPLA version uses the platform's central services and reviewed interaction bridge; students do not enter API keys.
Different activity forms
The examples represent three useful classes:
| Form | Primary student action | Typical pedagogical role |
|---|---|---|
| Phenomenon simulation | Vary parameters and compare outcomes | Prediction, model exploration, connecting representations |
| Procedural virtual lab | Follow, troubleshoot, and interpret a sequence | Experimental preparation and data reasoning |
| Hybrid workbench | Move between simulations, graphs, notes, documents, or questions | Coordinating several forms of physics knowledge |
Other AIPLA activities may use a conceptual dialogue, problem set, document, image submission, table, calculator, chart, checklist, or concept map without an embedded simulation.
Learning from AI mistakes
AI-generated explanations and illustrations can be convincing while still being physically wrong. AIPLA treats some failures as possible teaching material.
For example, image generators often depict a dust particle in front of a loudspeaker travelling away along a transverse sine-wave path. The plausible picture encodes a misconception: sound in air is longitudinal, and a nearby particle oscillates around its position rather than riding a drawn wave away from the speaker.
A teacher can ask students to compare the plausible image with a physical account, identify what the representation gets wrong, and explain how particle motion differs from the graph used to represent a wave. The value lies in the critique, not in presenting incorrect output without guidance.
What AIPLA adds to standalone artefacts
Compared with a standalone generated simulation or chatbot, the platform adds:
- a teacher-prepared learning context;
- a paired tutor that refers to the activity's actual representations;
- group-based student access without personal student accounts;
- central AI service configuration rather than browser-entered keys;
- reviewed and versioned interactive artefacts;
- visible sharing of relevant workbench interactions with the tutor;
- teacher review of activity use; and
- a path for approved research analysis.
Trying an activity
The development environment contains activities for demonstration and teacher review. Students participating in a class should use the group code and instructions supplied by their teacher.
Join a group or browse the teacher and student guides.
- Content status
- Current
- Maintained by
- AIPLA project team
- Last reviewed