Gains:
- Ability to use artificial intelligence as an idea diversification engine in layers of theme, narrative and principle rather than a single suggestion
- Being able to see the weak points before delivery by personally criticizing the concepts produced through budget, climate and user filters
- Ability to understand that idea generation is in artificial intelligence, selection and ownership is in humans, and transform the concept into measurable design principles.
The most difficult moment of any landscaping project is often the first: a blank page, an unclear plot and “what should happen here?” question. A concept is the central idea and narrative behind a design—the underlying idea that holds the project together, justifying each design decision (e.g., “a water line that returns the city-forgotten stream to the neighborhood”). A good concept guides all subsequent decisions (plant, material, circulation). In this unit, you will learn how to use artificial intelligence (AI) as the engine of concept generation; But you will see why the job of choosing, filtering and owning the idea is always up to you.
Why does the concept accelerate with AI?
The most valuable thing at the concept stage is variety: the more different approaches you see, the better your chances of finding the right one. The human mind tends to fixate on the first 2-3 ideas that come to mind (this is called "idea fixation"). AI, on the other hand, can suggest 10-15 different frames within minutes. Your job is not to produce them, but to extract and deepen the ones that fit your context.
Consider concept generation in three layers: (1) theme—the spirit of the project (“water memory,” “urban oasis,” “productive garden”); (2) narrative—how the theme will be put into place; (3) principles — clauses that translate the narrative into concrete design rules (“views to water from all points”, “hard ground ratio does not exceed 30%).” AI replicates ideas at all three layers.
Tip: Don't ask for a single "best concept" from the AI. Ask for “5 distinctly different themes.” Comparison always makes a better decision than a single suggestion.
Step by step concept generation
Step 1 — Nurture the context. Give the AI the land's climate, slope, environment, user, and budget constraint. The richer the context, the more localized the output.
Step 2 — Ask for theme variations. Ask him to produce 5-8 themes in different spirits; a one-sentence summary, strengths and weaknesses for each.
Step 3 — Select and deepen. Choose 2 themes you like and turn them into narrative and design principles.
Step 4 — Test it. I asked each concept "can it be done with this budget, does it make sense in this climate, who will use it?" Test it with questions. You can have the AI make this criticism yourself.
Step 5 — Turn it into narrative. Turn the chosen concept into a short, persuasive text that will explain it to the client/jury.
three mini cases
Case 1 — Breaking fixation. One office was always latching on to the "central square" idea for a college campus courtyard. AI was asked for 8 alternative themes; The "learning terrace" theme (the idea of turning the slope into a stepped seating area) came out. The team developed this idea and won first place in the competition. The idea came from AI, the design came from the team.
Case 2 — Narrative power. There was a technically good but "storyless" concept for a housing estate garden. YZ reframed the design with the “four-season walk” narrative; the same design appeared much stronger in presentation to the customer. The text was produced in 20 minutes, the architect corrected it in 10 minutes.
Case 3 — Early elimination. In a square project, AI produced 6 themes; The team filtered these through “maintenance cost” and “local climate.” 2 of the 6 themes were eliminated at the sketch stage because they required heavy watering. Early elimination prevented expensive overhaul later.
Four copyable templates
1) Theme variation:
Your role: conceptual landscape designer. Plot: [location, climate, slope, environment, user, budget level]. Produce me 6 design themes that are markedly different from each other. For each theme: one-sentence abstract, the main value it adds to the project, its strengths, and its weaknesses/risks. Don't name plants/ingredients yet.
2) Translating the concept into principles:
The theme I chose: [theme name and essence]. Turn this into 6-8 concrete design principles (e.g. hard ground ratio, view axes, circulation logic, shadow strategy). Let every principle be measurable or controllable.
3) Don't criticize the concept yourself:
Critique this concept with the eyes of a harsh jury: list the weak points in terms of budget realism, local climate suitability, maintenance burden, user realism and originality, and suggest improvements.
4) Presentation narrative:
Turn this concept into a compelling narrative text that can be told to investors in 90 seconds. Don't exaggerate and make promises that won't come true; Explain through concrete benefits and experience. Simplify technical terms.
Weak prompt / Strong prompt
Weak prompt:
Give me a creative concept for the park.
No context and no constraints; the output becomes cliché (“green oasis”) and alien to the locale.
Powerful prompt:
Your role: conceptual landscape designer. Land: semi-arid Central Anatolian climate, open to wind, 1.5 ha school garden; priority is shade, low water consumption and children's learning; Maintenance budget is limited. Produce 6 different themes; Write the essence, main values, strengths and weaknesses for each. Also mark themes with high water consumption. Don't name the plant.
Climate, priority, maintenance constraint and the "mark high water consumption" command make the output both creative and actionable.
Concept layers table
layer
Question
Contribution of AI
It's your decision
theme
What is the spirit of the project?
6-8 variations
Which spirit is right?
Narrative
How is the soul poured into space?
draft text
Decide on tone and accuracy
principles
What are the concrete rules?
Article suggestions
Measure and priority
test
Is it realistic?
List of criticisms
Eliminate/improve
presentation
How to explain?
persuasion text
Promise and fact check
Common mistakes
- Falling in love with the first idea of AI. Don't choose without comparing and asking for variation.
- Contextless prompt. If climate, budget and user are not given, the output will be cliché.
- Exaggeration in narrative. AI texts can be overly ambitious; Don't make promises that won't come true.
- Separating the concept from implementation. A beautiful but expensive/unkept concept is worthless; Eliminate early.
- Skipping the criticism step. Having yourself critique the concept allows you to spot weak points before delivery.
In summary
The real power of AI at the concept stage is idea triangulation: it frees you from fixating on the first 2-3 ideas. Move the process through the layers of theme, narrative, policy, testing, and presentation; Let the AI multiply in each layer, you filter it. Give rich context, filter the output through budget-climate-user, and have the concept personally critiqued before handing it over. The idea may come from AI; The choice, depth and ownership are yours.
Application task
Generate 6 themes for an existing (or imaginary) plot with the "Theme variation" template. Select two and turn them into concrete principles with the "Translating into principles" template. Then, test a concept you have chosen with the "self-criticism" template and fix at least three weak points.
checklist
- [ ] I created and compared at least 6 different themes.
- [ ] I richly added the context (climate, budget, user) to the prompt.
- [ ] I transformed the concept I chose into measurable principles.
- [ ] I filtered the concept through budget and climate.
- [ ] Before delivery, I criticized the concept and corrected the weak points.