NextGeneration AI-Design Alternatives to LEGO
Aimee Hadikrisno, FideliaAnastasya, Artem Muradyan (Tongji University)
Our design challenge was to re-design the Lego minifigure without infringingon their existing patents. Our key takeaway from this project was how to effectivelycollaborate with AI during brainstorming. We discovered that more detailed and specific prompts led to outputs that better aligned with our design goals. Through multiple iterations and refinements, we learned how to guide the AI to generate more relevant and useful ideas. Overall, this experience taught us how to use AI as a creative partner, enhancing our brainstorming process and
expanding our design possibilities.Brainstorming
We first used ChatGPT to help us write prompts to use to generateimages in Ideogram.ai (a text-to-image AI tool) to visualize our ideas. At each iteration, we further advanced our prompt as shown in Figure 1.1.
We also added visualizations from Ideogram.ai and pointed out toGPT which elements we liked or didn’t like to refine our prompt. Hence, the AI generated prompts were longer than the ones we first created, as shown in Figure 1.2. AI used more detailed descriptions and full sentences, while ours were shorter and mostly used single adjectives. We repeated this process until we got a satisfactory result, as seen in Figure 1.3.Prompts
We then used ideamap.ai (an AI-powered visual brainstorming platform) to generate mind maps based on the prompt. At each iteration, our team reviewed each idea and chose what we liked. We later incorporated those ideas into new prompts that we wrote manually by listing them at the end of the prompt to redefine the design direction with each iteration. (These preferences eventually made it into our finalized ideas list.) The process is demonstrated in Figures 2.1-2.5.HowAI was incorporated in the team
In our team process, everyone used AI programs together, allowingus to collaborate closely and share insights in real-time. We began by
collectively exploring different AI-generated ideas from tools like ChatGPT, Ideogram, and Adobe Firefly. Each team member participated in reviewing and discussing the AI content.
The decision-making was also a team effort, with everyonecontributing to evaluating the AI outputs and choosing which ideas to pursue.
Human user-input also helped with going in a specific direction and staying on the right track. This collaborative approach allowed us to stay aligned and adjust in every step of the designing process.Q& A
What did you do without AI?
- Defining the initial HMW questionwithout AI.
- Built and distributed thequestionnaire manually during the voting process.
What did AI add to this project and process?
- AI generates new ideas in thebrainstorming idea map process.
- AI can summarize and analyze userfeedback quickly.
- AI can make quick visualizations ofideas
What do we still need to know?
- We need more data to choose whichspecialized AI program are suitable for each task.
What could you do different?
- Write more specific prompts duringimage generation
- Be more specific and on point whendescribing our requirements to AI
- If an outcome of a certain processdoesn’t satisfy us, consider changing the tools and/or the entire workflow
Supported by the Fundamental Research Funds for the Central Universities
©2025