What is chain-of-thought prompting, and when does it help?
Chain-of-thought prompting asks the model to work through intermediate steps before giving the final answer. It tends to help on multi-step maths, logic and planning problems, but adds tokens, and the shown reasoning is not always a faithful account of how the answer was produced.
Explain it simply
In a maths exam, writing the working gets you to the right answer more often than jumping straight to it. The same holds for LLMs. Each generated step becomes context for the next one, so splitting a hard problem into small steps gives the model a better path.
You can trigger this with an instruction such as "think step by step", by showing examples with worked reasoning, or by using a model built to reason before answering.
It helps for arithmetic word problems, logic puzzles, and multi-constraint planning. It helps little for simple lookups and costs extra tokens and time. Also, the written steps can look convincing while being wrong or unrelated to the real computation, so check the final answer independently when stakes are high.
Example
Q: A train leaves at 9:40 and takes 2 hours 35 minutes. When does it arrive?
Steps: 9:40 + 2 hours = 11:40. 11:40 + 35 minutes = 12:15.
Answer: 12:15 PMWhat to say in the interview
"Chain-of-thought prompting makes the model produce intermediate reasoning before the answer, which helps multi-step problems. The trade-offs are more tokens and latency, and the reasoning text may not be a faithful explanation, so I verify outputs that matter."