简介
2026 arXiv paper on aligning LLM-generated interfaces with user tasks and preferences. The authors collect 720 UI-control preferences from 50 users across image-editing tasks, use that dataset to guide model reasoning, and evaluate generated UIs with 72 additional users.
内容总结
Why AlignUI Adds Value:
1. Preference alignment focus: Tackles how generated interfaces match user needs, not just prompt instructions
2. Dataset contribution: Uses a concrete preference dataset over UI controls and task contexts
3. User-study evidence: Evaluates outputs with a second group of 72 participants on unseen tasks
4. Design-method relevance: Bridges LLM UI generation with user-centered design evaluation
5. Collection fit: Complements personalization and GenUI theory papers with a practical alignment method
1. Preference alignment focus: Tackles how generated interfaces match user needs, not just prompt instructions
2. Dataset contribution: Uses a concrete preference dataset over UI controls and task contexts
3. User-study evidence: Evaluates outputs with a second group of 72 participants on unseen tasks
4. Design-method relevance: Bridges LLM UI generation with user-centered design evaluation
5. Collection fit: Complements personalization and GenUI theory papers with a practical alignment method
标签
preference-alignmentllm-generated-uihciuser-study