# Automated Item Generation
- [@gierlAutomaticItemGeneration2012] introduced theories and practices in AIG
field.
## Concepts
- Weak and Strong
- Weak Theory AIG: items are generated from models/templates, always vary only
names and numbers, never underlying logic. Mostly seeking to produce clones.
- Strong Theory AIG: adopt explicit cognitive models -- formal explanations of
how students reason through the task.
## Methods
- Item models/templates: "if a train leaves city X at \_\_ mph..."
- Item shells ("cloning"): "A patient presents with \[symptom set\]... what is
the most likely diagnosis?"
- Facet design: mapping out every possible way an item can vary in a domain.
- Concept based: generating items to test concept understanding.
- [[llm-assisted-aig|LLM assisted]]