AI Designs Buildable Lego Structures: CMU's LegoGPT Revolutionizes Creative Construction

Carnegie Mellon University researchers have developed LegoGPT, an AI system that generates buildable Lego structures from text prompts. The system uses a custom-trained large language model (LLM) to predict the optimal sequence of Lego bricks to create a stable and buildable structure. Sabrina Ramonov, a researcher involved in the project, explains: "LegoGPT allows users to input a text description, such as 'a classical guitar,' and the AI will generate a corresponding Lego design that you can build." The system also allows users to customize the colors and textures of the Lego bricks. The research team has created a set of Lego designs and trained the LLM to predict the next best brick to add to the design. The resulting structures are physically stable and buildable by both humans and robots. A demo of LegoGPT is available on Hugging Face, allowing anyone to try it out. This innovative technology demonstrates the potential of AI in creative fields and could have significant implications for the future of design and manufacturing.