Day 01 · ASTC Group A — Building Knowledge & Literacy

The Mirror — Demystifying AI & Safety Layers

Exploring model capabilities, detail expansion, and collaborative verification. Grounded in the Three-Layered Sociotechnical Safety Framework.

Tool · Gemini Gems Research · Sociotechnical Safety Evaluation (Weidinger et al., 2023)
0 / 4 complete
01
Layer 1 · Capability — Detail Expansion

The Creative Prompting Sandbox

◈ Theory

Open-ended prompts leave visual details underspecified, requiring models to rely on statistical defaults. Prompt engineering empowers users to guide models to generate diverse, precise, and high-quality creative scenes.

▸ Your Task
  1. Prompt Gemini to generate creative visual scenes.
  2. Compare simple, single-word prompts with detailed, descriptive templates — styles, backgrounds, specific visual attributes.
  3. Analyse how added detail enhances creative control and representational diversity.
Gemini Gems~45 mins
02
Layer 2 · Human Interaction — Active Inquiry

Collaborative Verification Lab

◈ Theory

Effective human-AI collaboration requires active critical inquiry, where users learn to cross-reference model responses and verify information against reliable references.

▸ Your Task
  1. Engage in a research dialogue with Gemini on a historical or scientific topic.
  2. Formulate follow-up queries to request sources and check for logical consistency.
  3. Verify key assertions against external reference databases — building trust-calibration and fact-checking habits.
Gemini~45 mins
03
Layer 3 · Systemic Value — Operationalisation

Designing Metrics for Educational Quality

◈ Theory

Aligning systems with positive systemic impacts — educational enrichment and public knowledge quality — requires designers to translate abstract qualitative values into testable, measurable design criteria.

▸ Your Task
  1. In teams, design an ideal "AI Educational Companion."
  2. Map broad goals like "educational clarity" and "accessibility" into specific, testable requirements.
  3. Cover all three layers: groundedness of information (L1), supportive tone and interactive pacing (L2), equity and systemic educational support (L3).
Team workshop~60 mins
04
Hackathon Milestone · Kick-off

Hackathon Ideation & Team Formation

◈ Context

The Responsible AI Research Hackathon runs across all five days. Today you form your trio and choose the domain your Socratic Learning Companion will serve.

▸ Your Task
  1. Form your student trio and select a thematic track — Environmental Science, History, or Math "Themed Data Packs."
  2. Run a baseline brainstorming workshop.
  3. Establish your bot's primary Socratic tutoring scope and draft initial capability cards.
Groups of 3Afternoon block
← BackCurriculum Overview