Problem definition means clearly describing a goal, its components, constraints, and relationships before asking AI to produce an answer. It is the planning stage before execution: instead of requesting a finished product, provide the structure that guides the solution. Subject-matter knowledge helps identify the right questions and requirements. section 4.7 section 4.0
Build a stronger AI request by specifying:
- The desired outcome and its workplace purpose. section 4.8
- Required tasks, inputs, dependencies, constraints, and performance needs. section 4.8 section 5.2
- The format, documentation, and level of detail expected in the result. section 4.8 section 5.2
From generic request to defined task
A basic request says: “Create a Python script that sorts a list of numbers in ascending order.” A defined request adds the data-analysis purpose, QuickSort or MergeSort, scalability for large datasets, inline comments, computational complexity, and reasons for the algorithm choice. The added context lets a knowledgeable user shape a result for a real need rather than accept generic output. section 4.8
Project documentation workflow
For an Oracle database migration, a senior project manager can ask AI to organize spreadsheet information into a project overview, task list, milestones, dependencies, resource allocations, communication plan, and risk register. Dependency and milestone flags become signals for the corresponding sections instead of remaining in the task list. section 5.2
Figure 4.1 Paint by numbers
Use the paint-by-numbers metaphor to see how a defined template gives AI boundaries while leaving room for variation inside them. section 4.7
Treat generated output as a starting point, not a guaranteed final product. Clear formulation improves relevance, but the requester still needs enough expertise to verify requirements, assumptions, and results. section 4.7 section 4.10
