#chain-of-thought

Question 1: Single choice

mediumAI-checkedLLM Fundamentals & Prompting

An LLM answers "Alice has 5 apples, throws away 3, gives 2 to Bob and Bob gives one back. How many does she have?" with 5. You add a similar solved example that writes out each calculation step (7 − 1 = 6, 6 − 4 = 2, 2 + 1 = 3) before the question. Why does this chain-of-thought prompt help?

It guarantees a correct answer on every arithmetic problem1
The model checks its own final answer against the example's final result2
The extra example shortens the prompt, which leaves more of the context window for the answer3
The model copies the worked method and computes each intermediate step before answering4

Timed test