Data/ML/AITable-stakesEmergingAround since 2022
Chain-of-Thought
Also written as CoT
A technique — and now a category of models, like OpenAI's o1 or Claude's extended thinking — where an AI works through intermediate reasoning steps before giving a final answer, improving accuracy on complex problems.
Think of it like
Like asking someone to show their work on a math problem instead of just blurting out an answer — walking through the steps out loud makes mistakes easier to catch.
Junior or senior?
Junior sounds like
Reaches for a reasoning model by default.
Senior sounds like
Can describe when they decided the extra latency and cost was actually worth it.
Ask them
“When did you decide a reasoning model was worth the extra latency and cost over a faster model?”