AI Guides › Prompting
By Nigel Guy · 2 min read
"Just be more specific" is the most common prompting advice going, and it's nearly useless on its own. It's true in the way "eat better" is true — nobody disagrees, and nobody is told which of the fifty things "better" could mean. So people respond by adding more words: more adjectives, more context, more sentences describing the vibe they're after. The prompt gets longer. The output barely changes, because length was never the problem.
The rule: "specific" isn't a volume of words, it's an answer to a named question — specific about the audience, the constraint, the format, or the comparison, not specific in general.
Vague specificity-advice fails because it doesn't say which of these you're missing. Pick the one that's actually loose in your prompt:
Before adding another descriptive word to a prompt, ask which of the four above you're actually short on. If the honest answer is "none of them, I just want it to be better," more words won't fix that — the prompt needs a decision made outside the prompt first, about what "better" means for this particular piece of work.
Skip stacking adjectives ("clear, concise, professional, engaging") as a substitute for naming a constraint — they don't compound, and a model reading four vague adjectives has no more information than one. And skip treating every prompt as under-specified; a one-off question you'll glance at once doesn't need this treatment, only the requests you're relying on.