What Is Reasoning in LLMs: Chain-of-Thought and Why Explicit Steps Help

~12 min read

LLMs generate one token at a time with a fixed amount of computation per token — 'reasoning' is what happens when a model is nudged to spend more of that computation on visible intermediate steps before committing to an answer.

What Is Reasoning in LLMs: Chain-of-Thought and Why Explicit Steps Help is a Pro topic

Sign in, then upgrade to Pro or Power to unlock this topic and the full AI Engineering curriculum.

Key points

  • An LLM predicting the next token immediately has a fixed, small computation budget for that single token — no built-in way to 'think harder' about a hard question
  • Generating explicit intermediate reasoning steps gives the model roughly N times the total computation before committing to a final answer, where N is the number of steps
  • Each reasoning step becomes context for the next, so the model literally conditions its next prediction on its own prior reasoning — the same mechanism as conditioning on any other context
  • This pattern is Chain-of-Thought (CoT), the simplest and most widely used reasoning technique — covered with book-grounded detail in this topic's Techniques subtopic
  • Explicit reasoning trades more generated tokens (more computation) for a better shot at correctness on tasks that genuinely benefit from multi-step derivation