How can artificial intelligence support landscape architects without replacing professional judgment? This question is explored in the master’s thesis by Ruben Haavik, completed in June 2026 as the final project of the Two-year Master’s degree in Global Landscape Architecture master program (GLA) at the Norwegian University of Life Sciences (NMBU). The thesis was supervised by Ramzi Hassan.
In his master thesis, Ruben Haavik investigates how AI can be evaluated in relation to the actual requirements of landscape architectural practice. Rather than asking whether AI is simply “useful” or “innovative,” the thesis takes a more precise approach: What professional tasks can AI support, and where must human judgment remain central?
The study develops a requirements-based framework for assessing AI in early-phase landscape architectural work. Four professional requirements are examined: user and site interpretation, conceptual design development, adaptive design reasoning, and critical design evaluation. These requirements are tested through a case study at Jens Bjelkes gate in Grünerløkka, Oslo.
The findings show a clear pattern. AI performed well in analytical and text-based tasks. It was able to identify stakeholders, summarize project goals, recognize conflicts and critically evaluate a design concept in a structured way. However, when the task required spatial production, the limitations became more visible. The AI could describe design ideas convincingly, but the accompanying plans contained spatial conflicts, overlapping elements and inconsistencies between text and drawing.
One of the most interesting findings is that AI can explain its reasoning in a professional and convincing way even when the output itself is weak. This raises an important concern for design practice: a fluent explanation is not the same as reliable professional judgment. The thesis therefore argues that AI outputs must be evaluated directly against clear professional criteria, rather than accepted because they appear well written or well reasoned.
The thesis is a valuable contribution to ongoing work at VR-Lab on AI, digital design methods and the future of landscape architectural practice. It shows that AI can be a useful support tool, especially for structuring information and reflecting on project material. At the same time, it confirms the need for careful professional control when AI is used in spatial and design-related tasks. The study points toward a responsible way of working with AI: define the professional task clearly, provide relevant project material, evaluate the output against explicit criteria, and keep the landscape architect responsible for interpretation, prioritization and final design decisions.