fCrit: A Visual Explanation System for Furniture Design Creative Support

AI explanations should match how creative professionals actually think and communicate. Rather than using generic technical explanations, fCrit tailors its critiques to users' own design language and cognitive framing, demonstrating that explainability in creative domains requires understanding domain-specific ways of thinking.

Dialogue-based interaction enables more responsible AI feedback in creative work. By using a conversational approach rather than one-way recommendations, fCrit allows designers to engage with AI reasoning iteratively, ask follow-up questions, and maintain creative agency throughout the design process.

Visual grounding is essential for trustworthy AI in visual creative fields. The system grounds its explanations in visual analysis rather than abstract reasoning, making it possible for users to verify AI critiques against what they actually see in their designs.

Structured knowledge bases can support more accountable AI reasoning in specialized domains. fCrit uses a formal design knowledge base to inform its multi-agent architecture, suggesting that transparent, domain-specific knowledge structures help make AI decision-making more auditable and less prone to unexplained outputs.

Human-centered explainability requires adapting to users' existing practices, not forcing them to adapt to AI. Rather than requiring designers to learn new terminology or frameworks to understand the system, fCrit demonstrates how responsible AI support should integrate into professionals' established workflows and communication styles.