Overly Vague Instructions

The most common beginner prompt mistake is vagueness: writing instructions so general that the model cannot reliably produce what you actually want.

"Write something good about AI."

This prompt contains no information about format, audience, length, tone, angle, or purpose. The model will produce something — but the chance that it matches what you had in mind is low. Vague prompts produce inconsistent, hard-to-predict outputs that require heavy editing.

A prompt is vague when it is missing one or more of the dimensions the model needs to make a good decision. The key dimensions:

• What — the specific topic or task
• Who — the target audience
• Why — the purpose or goal
• How — the format, structure, and style
• How much — the length or scope

You do not always need all five, but the more that are missing, the more the model must guess — and guesses introduce variability.

Before sending a prompt, run it through five diagnostic questions:

1. What specifically is being asked for? (not just "write about X" — what type of output?)
2. Who is the audience? (a subject matter expert? a beginner? a specific person?)
3. What is the output going to be used for?
4. What format and length does it need to be in?
5. Are there any constraints or things to avoid?

If you cannot answer most of these, the prompt is too vague. Spend 60 seconds adding the answers before sending.

Vague: "Tell me about machine learning."

Specific: "Write a 200-word explanation of what machine learning is, for a non-technical small business owner who is considering whether to use an AI tool for customer service. Use everyday language, no technical terms. Focus on what it does, not how it works internally."

The specific version gives the model everything it needs: topic, audience, length, vocabulary level, focus area. The output will be far more useful on the first try.

Many vague prompts suffer from an underspecified task — the type of output is not clear. "Write about climate change" could mean:

• A factual explainer article
• A persuasive essay
• A poem
• A social media post
• A business risk analysis
• A children's story

All of these are valid outputs for that prompt. When the task type is not specified, the model defaults to the most common pattern in its training data — usually a generic informational essay, which may not be what you need.

Audience is one of the most impactful dimensions in a prompt. The same topic should be explained completely differently for different audiences:

• "Explain neural networks to a machine learning engineer" → technical depth, specific terminology, code examples
• "Explain neural networks to a 12-year-old" → analogy-based, no equations, concrete examples
• "Explain neural networks to a marketing executive" → business impact, not mechanics, ROI focus

Without an audience specification, the model picks one arbitrarily. That choice is often wrong for your actual use case.

A systematic approach to turning vague prompts into specific ones: use the WHO-WHAT-FOR-HOW-HOW-MUCH framework to expand each dimension:

Certain words in prompts are almost always vague and should trigger a specificity check:

• Good — good by what standard? Who judges it good?
• Better — better than what?
• Appropriate — appropriate for what context or audience?
• Professional — professional in which industry? At which formality level?
• Improve — improve in what dimension? Clarity? Length? Accuracy?
• Interesting — interesting to whom?

When you catch these words in your own prompt, ask the follow-up question they imply. The answer becomes the specific instruction.

Specificity exists on a spectrum — you do not always need maximum specificity. The right level depends on the task:

• Creative exploration — lower specificity is intentional. Vague prompts produce surprising variety.
• Routine production — high specificity needed for consistency and reusability
• Factual retrieval — high specificity helps the model focus on the right aspect
• Brainstorming — medium specificity: define the domain but leave ideas open

Vagueness is a mistake when you need a specific output. It is a feature when you want creative range.

Sometimes the most efficient way to fix a vague prompt is to add an example of the output you want rather than trying to describe it in words:

"Write a product tagline in this style: 'Do more with less.' 'Think different.' 'Just do it.'"

The examples replace the need to define 'concise', 'punchy', 'brand voice', and 'memorable' in abstract terms. The model extracts the pattern from the examples more reliably than it infers it from descriptors. Examples are often the most efficient specificity fix available.

Over-specification is the opposite mistake from vagueness. A prompt with 15 highly specific constraints may produce output that technically satisfies all constraints but feels mechanical and wrong.

Calibrate by asking: "If I hand this prompt to someone who has never worked with me, what would they produce?" If a competent person could produce what you need from the prompt, it is specific enough. If they would need to ask clarifying questions, add the answers to the prompt. If they would feel over-constrained, remove some constraints.

Rank these four prompts from most vague to most specific:

(A) "Write about leadership."

(B) "Write a 300-word article about servant leadership for first-time managers at a tech startup, focusing on three specific behaviors they can practice in their first 30 days."

(C) "Write a leadership article that's good."

(D) "Write an article about leadership styles for managers."

Vagueness is the most common prompt mistake because it is easy to write vague prompts quickly. The cost is inconsistent, unpredictable output that requires heavy editing.

The fix is systematic: specify the WHO (audience), WHAT (task type), FOR (purpose), HOW (format and style), and HOW MUCH (length). Watch for common vague words (good, better, appropriate, professional) and replace them with specific answers to the question they imply. When description fails, use examples.

In the next lesson, you will tackle contradictory requirements — the opposite problem from vagueness.