Be CAREful: The Framework That Makes AI Useful
The Nielsen Norman Group’s new research reveals why your AI prompts aren’t working: You’re not being CAREful enough! The next time you interact with a generative AI tool, consider using:
The CARE Framework
- Context: Your role, project background, target users
- Ask: Specific, actionable requests (not vague asks)
- Rules: Clear constraints, style, quality standards
- Examples: Show what good (and bad) output looks like
Working with generative AI is similar to managing a new intern. Just as you wouldn’t tell an intern to “write a report,” you can’t expect AI to produce work that meets your expectations with vague prompts. Without proper context about your goals, specific instructions and examples of what success looks like, both your intern and your AI tool are likely to miss the mark. The quality of the output directly reflects the quality of the guidance you provide. Let’s look at an example.
Poor prompt: “Help me prepare for a meeting.”
CAREful prompt:
- Context: I’m presenting quarterly results to senior leadership who care about ROI.
- Ask: Create talking points for a 10-minute presentation on our marketing campaign performance.
- Rules: Lead with key metrics, include one challenge and solution, end with next steps.
- Examples: Insert examples of your past emails and writing that match the tone and voice of what you want AI to offer you.
CAREful prompts take more time upfront but deliver dramatically better results. For complex work where AI can genuinely accelerate your output, this investment pays off. If you’re interested in personalized training for your department, email Kenny Wilson at ken@purdue.edu to discuss options.
View this email in full here.