Emotion Prompting — Prompting Guide & Examples
Emotion prompting adds emotional context or stakes to a prompt to improve the quality and effort of AI responses. Research shows that phrases like 'This is very important to my career' or 'Take a deep breath and think carefully' can measurably improve model performance.
How It Works
Append emotional framing to your prompt that signals importance, urgency, or care. This doesn't make the AI 'feel' emotions, but it activates training patterns associated with high-quality, careful responses — similar to how a human would put in more effort on something important.
When to Use
Use emotion prompting for critical tasks where you want the model's best effort. Good for important emails, high-stakes analysis, creative work, and any task where you've noticed the model being lazy or superficial.
Model-Specific Tips
ChatGPT / GPT-4
GPT-4 responds to emotional context. Phrases like 'Take a deep breath' and 'This is critical' have been shown to improve accuracy in benchmarks.
Claude
Claude responds well to emotional framing. Expressing importance and care tends to produce more thorough, considered responses. Claude also responds to 'please think carefully.'
Gemini
Gemini responds to emotional prompting. Use phrases that signal importance and the need for careful consideration.
Pros & Cons
Pros
- ✓ Measurably improves response quality
- ✓ Dead simple to implement
- ✓ No additional tokens needed
- ✓ Works across all models and tasks
Cons
- ✗ Effects can be subtle
- ✗ Can feel manipulative or silly
- ✗ Not a substitute for good prompt structure
- ✗ Research results vary by task type
Example Prompts
This is extremely important for my startup's survival. We're pitching to investors tomorrow. Write the most compelling 60-second elevator pitch for our AI-powered legal document analyzer. Make every word count.
Take a deep breath and think carefully. I need you to review this contract clause for potential liabilities. Missing something could cost us millions. Be thorough and flag anything concerning, no matter how minor.
I'm counting on you for this. My team has been working on this feature for 3 months and we need perfect documentation. Write comprehensive API docs for these endpoints that a junior developer could follow without asking questions.