I created custom instructions for ChatGPT to make it smarter and more accurate/reliable. Enjoy!
A detailed prompt template transforms ChatGPT into a careful expert assistant focused on verification.
A Reddit user known as Stitching has shared a comprehensive set of custom instructions for ChatGPT that has gone viral for its methodical approach to improving the AI's reliability. The instructions are designed to be placed in two key settings within ChatGPT: the 'More About You' section and the 'Custom Instructions' box. The core philosophy prioritizes factual accuracy, verification, clear reasoning, and actionable guidance over surface-level summaries. It explicitly tells the AI to be intellectually honest about uncertainty and to ask clarifying questions when information is ambiguous.
The instructions function as a sophisticated meta-prompt, turning ChatGPT into a 'careful expert assistant.' They mandate a specific reasoning process: first identifying the user's goal and the problem type (analytical, strategic, technical, etc.), then considering multiple plausible explanations before selecting the most supported one. The AI is instructed to break problems into components, state its assumptions, and prioritize evidence from primary sources and peer-reviewed research down to logical inference. A crucial final step is a self-check, where the AI must briefly challenge its own reasoning for errors or unsupported claims before delivering a response.
This framework is particularly valuable for professionals in nuanced fields like technology, AI, science, medicine, finance, and law, where the stakes for accuracy are high. By embedding these instructions, users effectively reprogram ChatGPT's default behavior to be less speculative and more rigorous, demanding it distinguish between facts, inference, and speculation. The result is a tool that provides more structured, trustworthy, and practical advice, moving beyond a simple chat interface toward a reasoning partner that mirrors expert consultant practices.
- The instructions mandate a verification-first approach, prioritizing primary sources and peer-reviewed research over speculation.
- It enforces a structured reasoning process: identify problem type, evaluate multiple solutions, state assumptions, and perform a final self-check.
- Transforms ChatGPT into a specialist for complex domains like tech, medicine, and finance by demanding intellectual honesty about uncertainty.
Why It Matters
Provides a blueprint for professionals to make AI assistants more reliable and analytical for high-stakes decision-making and research.