despindler

Literacy	Description / Focus	Typical Artefact
Interaction Literacy: Designing Human–AI Dialogues	Structure instructions, roles, examples, and feedback loops to achieve predictable collaboration. Treat interaction as control, not mere instruction.	Collaboration Blueprint documenting roles, turn-taking, evaluation criteria, and revision cycles for a real task.
Mechanistic Literacy: Reasoning About Model Behaviour	Build intuition for token prediction, attention, context effects, and decoding settings. Link input changes to behavioural outcomes.	Model Behaviour Map showing experiments across prompts/temperatures/length penalties and the observed effects.
Data Literacy: Curating and Testing Knowledge Grounding	Ground outputs in vetted sources; design retrieval/citation; measure factual reliability and bias.	Grounding Protocol that compares unguided vs. retrieved/cited outputs with a small evaluation.
**S

despindler

#	Perspective	What It Focuses On	What It Can Explain / Predict	What It Cannot Explain (Limitation)	Typical Blind Spot

despindler

#	Project	Primary Perspective(s) / Literacy Focus	Learning Focus / Capability	Why It’s Future-Proof	Typical Artefact
1	Interaction Literacy: Talking to Machines with Purpose	Interaction	Structure prompts as purposeful dialogues: define roles, intentions, and evaluation criteria. Build clarity, reflection, and feedback loops.	Language-based interaction will remain universal across AI systems.	Collaboration Blueprint – evolving conversation workflow for a personal or professional task.
2	Mechanistic Curiosity: Understanding Model Behaviour	Mechanistic	Experiment systematically with parameters (temperature, context length, decoding). Build intuition for probabilistic generation.	Core model dynamics persist across architectures.	Behaviour Map – short report visualizing cause–effect relationships between input changes and