The Three Wishes Problem: Avoiding AI Design Pitfalls
AI agents can deliver technically correct but useless results. Avoid the three wishes problem through intent clarification and early validation.
1 min read
AI designagent designAI pitfallsspecification problemsAI UX
Like genie wishes gone wrong, AI agents can produce technically correct but practically useless results. Understanding this "three wishes problem" improves agent design.
The Problem Pattern
Users ask for what they think they want. AI delivers literally. The result doesn't actually solve the underlying need. Blame gets assigned to AI when the real issue is specification.
Design Solutions
- Clarify intent: Understand the underlying goal, not just the stated request
- Validate assumptions: Surface and confirm implicit expectations
- Iterate quickly: Get feedback before significant investment
Practical Application
Design agents that ask clarifying questions. Build in checkpoints. Create feedback loops that surface misalignment early.