Clear Instructions

Why Clarity Matters

Clarity is the single most important property of a prompt. The model interprets your instructions literally and fills in ambiguity with statistical defaults — which may not match what you actually wanted.

The Core Problem

When you write "Tell me about Python," you know what you mean. The model does not. It has to guess: beginner or expert? Overview or deep dive? History, syntax, or use cases? Without clear instructions, it produces the average of all possible answers — useful to nobody in particular.

How the Model Handles Ambiguity

Silent Resolution

The model does not ask for clarification. It does not flag uncertainty about your intent. It simply generates the most statistically likely continuation of your prompt. Ambiguity is resolved silently — always in a direction the model chooses, not you. The fix is always the same: be more specific.

Being Specific

Specific instructions answer the questions the model would need to ask if it could. Every question you answer in the prompt is one fewer assumption the model makes.

01

What exactly should the output contain? — name the subject and the specific aspect you care about

02

For whom? — specify the audience: a beginner, a senior engineer, a non-technical manager, a child

03

In what format? — list, table, JSON, code block, paragraph, numbered steps

04

At what length? — one sentence, five bullet points, under 100 words, one paragraph

05

With what scope? — what topics to include, and explicitly what to exclude

Common Vagueness Patterns

01

Open-ended verbs — "Write something about X" — "something" gives the model complete freedom to decide what to produce

02

Undefined audience — "Explain machine learning" — at what level? Assumes you know what the reader needs, but the model doesn't

03

Undefined length — "Give me a summary" — one sentence or two pages? Both are technically summaries

04

Undefined format — "List the pros and cons" — bullet list? Table? Flowing prose? Models vary in their defaults

05

Undefined scope — "Tell me everything about Python" — this could fill a book; the model has to decide what to include and cut

Before and After

Vague

"Explain neural networks."

No audience, no format, no length, no scope. The model writes a generic multi-paragraph overview at an assumed reading level.

"Summarize this."

No length target. The model may produce one sentence or ten. No format. No indication of what aspects matter.

Specific

"In exactly 3 sentences, explain neural networks to someone who understands basic math but has never studied AI."

Audience defined. Length defined. Reading level implied. The model has a clear target.

"Summarize the key findings in 5 bullet points. Focus on business impact, not technical detail."

Format defined. Length defined. Scope defined. Focus defined.

Rules for Clear Instructions

01

Start with an action verb — write, summarize, classify, extract, explain, translate, fix, compare

02

Name the subject explicitly — do not use pronouns or references the model cannot resolve without context

03

Specify the audience — "for a beginner," "for a senior engineer," "for a non-technical stakeholder"

04

Set a length target — "in one sentence," "in under 100 words," "in exactly 5 bullet points"

05

Name the format — "as a JSON object," "as a markdown table," "in plain prose with no headers"

06

Set the scope — tell the model what to focus on and explicitly what to exclude

Summary

Key Takeaways
  • Clarity is the most important property of a prompt — ambiguity is resolved silently by the model
  • The model cannot ask for clarification — it fills gaps with statistical defaults
  • Specific instructions answer: what to produce, for whom, in what format, at what length, with what scope
  • Common vagueness patterns: open-ended verbs, undefined audience, undefined length, format, and scope
  • Rules for clear instructions: action verb, explicit subject, audience, length, format, and scope
  • Every element you specify is one fewer assumption the model makes

Need Help?

Ask the AI if you need help understanding why vague prompts fail, how to write specific instructions, how to identify and fix common clarity problems, or how to structure a prompt for a specific task you have in mind.