Better Instructions
Clear instructions reduce ambiguity by telling the model what outcome matters and what boundaries it must respect.
Clear instructions reduce ambiguity by telling the model what outcome matters and what boundaries it must respect.
What should be accomplished?
What should the model know?
What should the result look like?
Understand the working model
A useful instruction usually combines a concrete goal, relevant context, constraints and a clear output format. You do not need “magic words”; you need enough structure for the task to be unambiguous.
If the first result is weak, diagnose why before adding more text. Was the goal vague? Was context missing? Was the output format undefined? Improve the missing part instead of randomly making the prompt longer.
Practice
Apply the pattern above to one real task you already do. Keep the first version small enough that you can inspect every input and output.
Do not rely on vague verbs like “improve” or “make it better” without defining what better means.
Good instructions make the desired result observable: goal, context, constraints and output format.