Part II · Chapter 19

19. Structured chains of thought

Major question

How can you communicate to AI the structure of your thinking?

Description

A chain of thought is a first way to share with AI how you think. Instead of asking only for an outcome, you make the order of your intellectual work explicit: hypothesis, literature review, arguments for, arguments against, then drafting. The path becomes inspectable - for you and for the model - without claiming access to private token-by-token internal reasoning.

Reformulated prompt

Analyse the economic value of beauty in tourism by following these steps strictly: 1) Specify the central hypothesis; 2) Conduct a literature review; 3) Synthesise the arguments for; 4) Synthesise the arguments against; 5) Write a 3,000-character article, based only on relationships and arguments that can be reasonably defended.

Expected impact

You no longer delegate only the outcome: you first share the structure of your intention. The model must follow your chain (hypothesis, literature, for, against, drafting) and make each step visible before the 3,000-character article.

Way of thinking -> explicit chain -> shared execution -> inspectable artefact.

  • Accuracy note: Do not equate a displayed reasoning structure with complete access to the model's internal computation.