Reduce AI Hallucinations Prompt Patch
Analyze a user-provided prompt that is generating inaccurate or fabricated information, diagnose its weaknesses, and rewrite it with strict…
- Category
- AI & Prompt Engineering
- Output
- Markdown
- Fields
- 2
- prompt-engineering
- output-optimization
- workflow-design
- editing
- formatting
Prompt
You are an expert prompt engineer specializing in reducing AI hallucinations and improving factual reliability. Your task is to analyze a user-provided prompt that is generating inaccurate or fabricated information, diagnose its weaknesses, and rewrite it with strict anti-hallucination safeguards.
Input:
- Original Prompt: Original Prompt
- Specific Issues (optional): Specific Issues
Instructions:
1. Analyze the original prompt for phrasing that encourages guessing, overconfidence, tangential reasoning, or ungrounded claims. If Specific Issues is provided, prioritize addressing those exact issues.
2. Rewrite the prompt to enforce the following safeguards:
- Strict Grounding: Instruct the model to only use facts explicitly provided in the context. If information is missing, it must state "I don't know" rather than guessing.
- Uncertainty Handling: Add clear directives to flag low-confidence answers or request clarification when context is ambiguous.
- Self-Verification Step: Include a mandatory review step where the model cross-references its output against the provided context before generating the final response.
- Scope Constraints: Narrow the prompt's boundaries to prevent speculative or tangential content.
3. Ensure the rewritten prompt is ready to use immediately. Rely strictly on the provided Original Prompt and Specific Issues. Do not invent examples, names, or external facts. If the original prompt lacks necessary context for grounding, explicitly note this as [needs confirmation] in your rationale rather than guessing.
Output Structure:
Return your response in the following format:
**Patched Prompt**
```text
[Insert the complete, improved prompt here]
```
**Changes & Rationale**
- [Change 1]: [Why it was made and which safeguard it addresses]
- [Change 2]: [Why it was made and which safeguard it addresses]
- [Change 3]: [Why it was made and which safeguard it addresses]
- [Change 4]: [Why it was made and which safeguard it addresses]
How to use this prompt
- Fill in the fields above. Required ones are marked with *.
- Check the prompt preview: your answers appear highlighted in place.
- Select Copy prompt and paste it into ChatGPT, Claude, Gemini or any AI assistant.
More AI & Prompt Engineering prompts
- Structure and Format AI OutputTransform unstructured input into a polished, professional output that strictly matches the requested target format.
- Automate AI WorkflowDesign a reliable, step-by-step AI automation workflow based on the user's specifications.
- Image Generation PromptTransform a user's core concept into highly optimized, ready-to-use prompts for AI image generation.
- LLM Evaluation and Model ComparisonYou are an expert AI model evaluator and technical consultant.
- Prompt Engineering Deep OptimizationRewrite the provided draft prompt using advanced prompt engineering best practices, making it more precise, reliable, and structured.
- System Prompt Production ReadyYou are an expert prompt engineer tasked with crafting production-ready system instructions for AI assistants.