04. Week 12 — Daily Recall

Monday

1. The quick guess vs the thinking pause — define both in one sentence each.
2. Why can a fluent answer still be globally wrong?
3. [W11] Verifier vs judge — when would you use each?

Tuesday

1. Zero-shot CoT vs few-shot CoT — difference and use case.
2. Self-consistency — what does it add beyond a single chain?
3. Give one example where CoT helps and one where it hurts.

Wednesday

1. Hidden CoT — why do model providers prefer it?
2. Test-time compute scaling — what does "thinking longer" mean mechanically?
3. Search + verification + backtracking — explain the loop from memory.

Thursday

1. Cost-quality-latency triangle — state the trade-off clearly.
2. Why is routing a Lead-level product decision, not only a prompt tweak?
3. Draw the cheap-first escalation pattern from memory.

Friday

1. Faithfulness — why can a correct answer still have bad reasoning?
2. Benchmark gaming — name 2 ways reasoning evals can mislead.
3. What 4 concepts from this module does Module 13 assume you already own?