Redditor Discovers One-Word Prompt Technique to Significantly Improve ChatGPT Response Quality

03.02.2026

A Reddit user has discovered an effective prompt engineering technique that dramatically improves ChatGPT's output quality using a single-word feedback approach. Instead of providing detailed critiques, the user simply responds with "mid" (slang for mediocre) to subpar responses from the chatbot.

This minimalist feedback strategy triggers a notable behavioral change in ChatGPT, prompting the AI to immediately regenerate and improve its response rather than offering explanations or justifications for the initial output. The technique essentially bypasses the model's tendency to defend or explain its answers, forcing it directly into iteration mode.

According to the original post, alternative single-word prompts that yield similar results include:

• "boring"
• "cringe"
• "you're like an NPC"

Independent verification confirms that this prompt engineering method demonstrates measurable effectiveness in practice, representing a simple yet powerful approach to optimizing AI-generated content through strategic negative feedback.

Source: https://www.reddit.com/r/PromptEngineering/comments/1qsf4j3/i_started_replying_mid_to_chatgpts_responses_and/

Tags: Generative AI ChatGPT Prompt Engineering LLM AI Optimization

Share: VK Telegram Twitter