Skip to content

Day 4 Exercises

Practice advanced optimisation, conversational learning, and ethical AI use.

Exercise 1: Calculating Your AI Footprint

Part A: Estimation

Calculate the environmental impact of a typical literature review task (50 complex queries).

  1. Water Consumption: 50 queries × 50 mL = _ mL (approx. _ cups of water)
  2. Carbon Footprint: 50 queries × 3g CO₂ = _ g CO₂ (approx. _ km driven in a car)

Part B: The Impact of Optimization

If you apply the One-Shot Principle and reduce the task from 50 queries to 10 queries:

  1. New Carbon Cost: 10 queries × 3g CO₂ = ____ g CO₂
  2. Savings: _ g CO₂ (an _% reduction)

Exercise 2: Workflow Optimisation Challenge

Your Task

Identify and optimise your most inefficient AI workflow.

Step 1: Document Current State - Task description: __ - Queries per instance: ___ - Time: ___ - Pain points: __

Step 2: Apply Optimisation Strategies Choose 3+ strategies from Day 4 content (e.g., One-Shot, Context Pruning, Batching). 1. Strategy: __ 2. Strategy: __ 3. Strategy: _____

Step 3: Test Optimised Approach Execute your optimised workflow in the sandbox. - New Query Count: ___ (Reduction: %) - New Time: ___ (Reduction: %) - Quality: Same / Better / Worse

Exercise 3: Deep "Why" Dialogue

Scenario

You need to improve the mechanical properties of your PLA/graphene composite.

Your Task

Design and execute a 10-question "why" cascade that takes you from surface understanding to expert-level insight.

Starting point: 

User: "How do I improve mechanical properties of PLA/graphene composite?"
AI: "Optimise graphene dispersion and interfacial adhesion"

Your "why" cascade: 1. Why is dispersion important for mechanical properties? 2. __ 3. _ 4. __ 5. _ ... (up to 10)

Exercise 4: Bias Detection Challenge

Part A: Identify Bias

Review this AI-generated content for bias:

Professor John Smith of MIT and Dr. Michael Brown of Oxford have pioneered techniques that are now standard practice. Industry positions typically require networking at the annual MRS Fall Meeting in Boston. Senior male researchers often provide valuable guidance to female junior colleagues.

Biases identified: 1. Type: __ (Evidence: _) 2. Type: __ (Evidence: _)

Exercise 5: Privacy Risk Assessment

For each scenario, determine the appropriate action (Public AI vs. Local Sandbox vs. Never):

  1. Scenario: Summarising 10 published papers from Nature.
    • Action: _____
  2. Scenario: Formatting a lab note for experiment #2847 (patent-pending).
    • Action: _____
  3. Scenario: Drafting a generic template for a safety protocol.
    • Action: _____
  4. Scenario: Comparing your unpublished synthesis method to competitor patents.
    • Action: _____

Next: Wrap-Up & Certification →