Ever wondered why your sleek, AI-powered design tool sometimes spits out nonsensical visuals? The problem lies in how AI handles design contradictions—those creative clashes that can spark innovation in human designers but often stump machines. Imagine asking an AI to create a "round bar chart." A bar chart's purpose is to compare data through the lengths of rectangular bars. Making it round defeats that purpose. This seemingly simple contradiction highlights a fundamental limitation in current AI design tools: they prioritize similarity and established patterns over novel or divergent thinking. Large language models (LLMs), the brains behind many AI design tools, struggle with these creative leaps. They excel at generating outputs similar to what they've been trained on, but falter when faced with conflicting design elements. This is where human creativity comes in. We can imagine and create visuals that defy traditional rules. This research challenges us to rethink how we approach AI-driven design. Instead of simply feeding instructions into a machine, we need to find ways to bridge the gap between human creativity and AI's computational power. How can we teach AI to embrace the unexpected and generate truly innovative visuals? This is a crucial question for the future of design, as we explore the exciting potential of AI-powered creativity, but also recognize the crucial role of human ingenuity in guiding this powerful technology.