Chain of Thought
Chain of thought (CoT) is a prompting approach where the model shows intermediate reasoning steps before the final answer.
For designers, CoT is less about math tricks and more about whether users should see reasoning, summaries, or only conclusions.
Share
What it means
The model is instructed to think step-by-step (often internally or visibly), improving accuracy on multi-step tasks at the cost of length and latency.
Why designers should care
Expose reasoning when trust and audit matter; hide or collapse it when users need speed. Always offer a concise final answer layer.
Example
A prioritization assistant expands “Why this ranking?” to show factors (impact, effort, dependencies), but the default view is a sorted list with one-line rationales.
Common mistakes
- • Dumping raw chain-of-thought walls on users who asked for a decision.
- • Hiding reasoning entirely on high-stakes recommendations with no audit trail.
- • Confusing verbose output with higher quality without verification UI.