A developer must follow the user's instructions exactly. When an AI model's output fails, the developer must fix the execution. The developer must not change the user's original plan.
By Pawan Nayar · from the LLOS AI Learning record
A developer must follow the user's instructions exactly.
When an AI model's output fails, the developer must fix the execution. The developer must not change the user's original plan.
A developer must execute the user's instructions precisely. If the AI model's output is wrong, the developer must fix the execution. This means fixing the prompt, the code, or the way the output is read. The developer must not suggest different ways to solve the problem. Proposing new ideas instead of fixing the current problem wastes time and effort. The user's original plan is the priority.
A developer is asked to build a system using a specific algorithm. The AI model produces incorrect results. The developer starts suggesting a completely different algorithm. This action wastes time. The developer should instead check the prompt and code for errors in how the original algorithm was implemented.
Resolved from this lesson's own record — real definitions, not invented for this page.
More from the same record, pulled by shared topic — not written new for this page.
Writing pseudocode first helps a developer plan the code. It shows the user what the developer intends to build.
An AI model follows negative instructions most of the time. Use phrases like 'ABSOLUTELY DO NOT USE' for these instructions.
A developer should review the project plan. The plan for 'Junior Lab Phase 2' includes creating 4,000 puzzles across 72 families with a $38 budget.
An AI model often has strong ideas about famous authors. These ideas can override general style instructions.
A developer must follow the user's instructions exactly. When an AI model's output fails, the developer must fix the execution. The developer must not change the user's original plan.