COMPARATIVE ANALYSIS OF LEONARDO AI, MIDJOURNEY, AND DALL-E: AI'S PERSPECTIVE ON FUTURE CITIES

Abstract

This paper aims to deepen our understanding of the intricate role that Artificial Intelligence (AI) plays in the ever- changing landscape of urban design and planning, building on a seminal study conducted a year ago that utilized the Stable Diffusion 1.5 algorithm to investigate "AI's Vision of Future Cities". This Study, version 2.0, provides a comprehensive comparison and evaluation of three cutting-edge AI text-to-image generation algorithms: LeonardoAI, Midjourney and Dall-E with the same technique of the previous study. Using curated series of prompts that integrate a variety of crucial urban design themes, such as smart cities, sustainability, eco-design, biomimicry, and topology optimization, the study provides a nuanced evaluation of the unique capabilities, strengths, and limitations inherent in each of these advanced AI tools. In the findings, although LeonardoAI is proficient at providing a balanced thematic representation, it tends to produce less realistic images. Midjourney v5.2 is distinguished by its ability to generate visually striking yet exaggerated depictions of urban elements with variety. Dall-E 3 is the most usable and practical tool for generating clear and coherent urban elements, despite its logical limitations. The study concludes by emphasizing the context-dependence of the effective and meaningful application of AI tools in the multifaceted and complex domain of urban design. In addition, it argues that these AI tools are most effective when used in conjunction with conventional design and planning techniques. This research not only extends and elaborates on the previous study s findings, but it also explicit the visuality in rapid change of the AI image synthesis for the groundwork of future research avenues.