📋 Full Prompt

Evaluate two AI responses: [Response A] and [Response B]. To avoid structural biases in LLM evaluation (e.g., favoring the first presented response, or a longer response), strictly follow this procedure: 1. Extract and compare only the "facts" from each response, excluding length or formatting. 2. Review [Response B] first, then review [Response A]. 3. Declare the final unbiased winner (A, B, or Tie) and explain the reason based purely on factual comparison in 2 sentences.

💬 Input Example

Response A (Long, inaccurate), Response B (Short, core is accurate)

✨ Output Example

Analysis: [Response B] wins on specificity. [Response B] accurately delivered the core facts, whereas [Response A] was merely long and contained errors. Conclusion: B wins.

🏷️ Tags

#bias#prompt evaluation#AB test#harness

💡 Usage Tips

✅Replace the variables in brackets ([]) with your specific situation.
✅Specify the desired tone and length of writing for more accurate results.
✅Request additional modifications based on the generated output to improve quality.
✅Specifying a particular audience allows AI to adjust vocabulary and explanation level.

🔗 Related Prompts

LLM-as-a-Judge Evaluator

Fairly evaluates the output results of other prompts and assigns a score from 1 to 5 based on a provided rubric.

Domain Edge Case Generator

Automatically generates difficult and ambiguous edge case scenarios to test how well a specific system can respond.

Red Teaming Vulnerability Attacker

Automatically generates adversarial roleplay scenarios to bypass AI safety policies or induce hallucinations for system security testing.

Effective Writing Strategies with AI

AI writing prompts let you efficiently produce blog posts, business emails, press releases, novels, and more. The key is to assign AI a precise role, and clearly specify the desired tone, length, and target audience. For example, requesting "write a 2000-word SEO-optimized blog post in a professional tone" yields significantly higher-quality output. After generating a draft, iterative feedback helps elevate the final quality step by step.

This prompt belongs to the Harness Engineering category, optimized for All at a Advanced level. It is systematically designed applying core prompt engineering principles: role assignment, context setting, and constraints.

Explore More Prompts

Discover 90+ verified prompts in the PromptGenie library.

Browse Library