overprompting
/oh-vər-PROMPT-ing/·noun & verb
Micromanaging an AI with excessive instructions — often getting worse results, slower.
Definition
Prompting practice. Overloading a model with redundant constraints, re-explanations and corrections until output quality degrades — the machine equivalent of backseat driving.
Behavioral. Reflexively prompting an AI for tasks one could finish faster alone; outsourcing thought past the point of return.
Origin
First recorded 2026, as AI assistants became coworkers and users discovered that trust, not verbosity, is a prompting skill.
Usage
- Forty-line prompt, three paragraphs of apologies for the prompt — classic overprompting.
- She wasn't stuck on the problem; she was stuck overprompting it.
- The fix for overprompting is usually a shorter brief and the courage to press enter.
Frequently asked
How do I know I'm overprompting?
Your prompts grow while results flatten; you re-specify things the model already did correctly; drafting the prompt takes longer than doing the task.
Is detailed prompting always bad?
No. Rich context helps complex work. Overprompting is redundant or contradictory detail — instructions fighting each other.
What's the opposite of overprompting?
A clear brief, sensible defaults, and reviewing outcomes instead of dictating keystrokes — delegation rather than dictation.
Is overprompting a productivity problem or a psychology problem?
Both. It starts as technique and hardens into habit: the reflex to route every thought through a model.