How Are Landscape Architects Thinking About the Future of AI?

The 2023 Patrick T. Curran Fellowship

From tech across virtually every sector, one topic seized attention in 2023: artificial intelligence. This year, SWA took a big swing, supporting eight researchers to study how AI will impact design through our annual Patrick T. Curran Fellowship. Here’s what we found.

You’ve probably used AI today. Every time you unlock your phone with facial recognition, map a route, ask a smart speaker a question, or open an app—banking, shopping, streaming, social media—AI is there in the background, powering the system. Quietly, it has slipped into our lives and browsers in countless ways.

What changed? In 2023, fueled by a series of breakthroughs after the public release of OpenAI’s ChatGPT, AI has shifted to center stage across virtually every industry, spurring curiosity, apprehension, and sweeping regulatory probes with uncertain outcomes. This fall, its governance issues splashed across every major publication in the U.S.

Many industries quickly followed suit. In architecture, firms moved to study and deploy a new range of image-making tools like Dalle-E, Midjourney, and Stable Diffusion. For a sector in economic tumult, known for hours of design iteration, the promise was high, at least in theory—cutting-edge software to lower costs, accelerate production, and heighten creativity. In turn, this has been balanced with an equal amount of open questions—questions of labor, competition, and authorship for a service-based model in transformation.

Landscapes, of course, are a different prospect than buildings. Composed of living organisms and materials set into earth, they’re subject to dynamic environmental factors—climate, soil, topography, and site context—and molded by complex social systems. Could generative AI grasp network-based thinking? Despite early challenges, could design teams benefit? With its toolset changing by the day, what could (or should) be the guardrails? What can we learn from other industries grappling with the same unknowns?

This fall, eight SWA designers across four teams carefully examined applications of these technologies for landscape architects, planners, and urban designers, focused primarily on the combined use of generative AI and large language models (LLMs) for design research, visual production, and site analysis. Then, in November, we came together to talk about it—and what comes next.

Artificially Natural: A Playbook for AI-Enabled Landscape Design

Liqiu Xu

At its most basic level, design work is defined by phases. In her research, Liqiu Xu examined AI applications by phase, outlining tailored uses, shortfalls, and areas for future research in each. Specifically, Liqiu looked at the individual or combined use of 15 different tools including chatbots, image and video generators, and 2D-to-3D conversion tools, proposing potential workflow integrations at each step.
Tools studied: ChatGPT, Claude, Google Bard, Llama, Midjourney, Dall-E, Stable Diffusion, Adobe Firefly, Runway Gen-2, Leiapix, CSM, Depth Map, Veras, PromeAI & Autodesk Fusion 360

AI in Landscape Architecture

Shi Chen, Yaxin Cao, Wenjing Fang & Yeqing “Jerry” Shang

At a larger scale, the second team studied the question of potential firmwide integration of AI systems, tackling a focused suite of three tools in more depth. As an outcome, the team developed a series of proprietary models trained on SWA content—testing the utility of each software, proposing guidelines, and suggesting an operational strategy for how these could be used collaboratively across SWA’s eight studios.
Tools studied: Midjourney, Stable Diffusion, ChatGPT Plus (for Grasshopper)

Generative Nature

Nandi Yang, Xun Liu (USC), Western EcoSystems Technology

The third team zoomed into a single topic: ecological design. In partnership with WEST, the team created and validated a tool to generate unique planting models through a combination of human expertise, parametric tools, statistical AI, and generative AI. They then tested it against a variety of factors, including scale (thousands of acres to under a single acre) and project type (whether reforestation or afforestation). At a broader scale, the team proposed further research to look at opportunities in the conservation world—especially on large-scale projects like the Great Green Wall Initiative or Amazon reforestation.
Tools studied: Midjourney, Pix2Pix

Capabilities of AI for Future Design

Tamaki Inahata

Focusing on the client side, Tamaki researched potential applications of AI in future retail spaces, transportation infrastructure, and workplaces. Through a series of speculative projects, her research illustrated how the very foundation of these projects might change as AI expands toward universal use, asking questions like: how will self-driving vehicles change streetscape design? As retail steers away from traditional sales floors to experiential spaces, is there an expanded role for landscape? Finally, as virtual collaboration tools evolve, what’s the future of office design?

From macro to micro, each team shared a common baseline: adapting in real time to change, often incorporating new plugins, softwares, and peer research the same week of release. If 2023 is any indication, 2024 and beyond will spark change on an even larger scale, affecting the AEC industry in ways we are only just beginning to understand.

Two additional 2023 fellows, Weston Henry and Mariana Ricker, examined issues unrelated to AI but equally important: biodiversity loss and decarbonization, respectively. Stay tuned for updates related to their research, and the AI teams, in the coming weeks.

Principal Advisors
Gerdo Aquino
Shuyi Chang
Hui-Li Lee
Sean O’Malley
David Thompson
Shuntaro Yahiro
Rene Bihan
Ying-yu Hung
Lisong Huang (external)

Additional Support
Anya Domlesky
Michael Robinson
Jonah Susskind