Gains:
- Ability to describe the logic of urban growth, sprawl and scenario simulation and the role of AI in comparison with scenario fiction
- Ability to transparently set up and compare different growth scenarios with input assumptions, constraints and indicator sets
- Ability to understand that the simulation output is a scenario, not a prediction, and that a formal planning process and expert evaluation are required for the decision.
A city grows; But where, how much and how to grow is a choice. One of the most difficult tasks for a planner is to compare the questions "if we continue like this" and "if we implement this policy" side by side. The tool that does this is urban growth simulation: an approach that uses past growth patterns, land suitability, and constraints to model where future development might spread. Common techniques include cellular automata and growth probability models. AI is a powerful aid in setting up the scenarios of these simulations, adjusting input assumptions, comparing results with indicators and reporting. But the critical point is this: simulation is a scenario, not a prediction. "Under the following assumptions, this could happen," he says; He doesn't say "it will be like this". The decision is made through the formal planning process and expert evaluation.
How simulation works
Typical inputs of an urban growth simulation: existing construction, land suitability (slope, flood, protection), transport accessibility, population/demand pressure and constraints (areas closed to construction: forest, agricultural protection, disaster area). Based on these inputs, the model assigns a "growth probability" to each area and distributes the demand to the space according to this probability. The result is the possible future construction pattern.
Each entry here carries an assumption: how much will demand increase, which areas will be protected, how will transportation develop? Different assumptions produce completely different futures. That's why not a single simulation, but several scenarios are run and compared: "uncontrolled expansion", "compact development", "protection priority", etc.
Caution: A simulation map looks very convincing; colourful, detailed, scientific. But this image is not a reality, but the result of a set of assumptions. Change the assumptions, the map changes. Never present simulation as “this is the future”; Present it as "under these assumptions this could happen".
The public value of simulation is not to "know" the future, but to make the costs of different futures visible today. Seeing on a map in advance which agricultural basin, which water source, and which forest belt a city will lose if it expands uncontrollably offers a concrete choice to the decision maker. That's why a good simulation study asks "what is the most likely future?" not to the question "what future do we want and what should we do for it today?" It serves the question. To feed into this question, AI rapidly replicates scenarios and makes the results of each comparable; But which future is preferred is not a technical outcome, but a democratic decision. Positioning the simulation as a basis for discussion rather than a prophecy also protects it from being a tool of manipulation.
Tip: When presenting a scenario to parliament, always include at least one “undesirable” scenario (e.g. uncontrolled spread). Presenting only the preferred scenario actually closes down the discussion while it appears there is a choice.
Another common misuse of simulation output is to translate results for a very distant future (e.g. 50-100 years) into a definitive planning decision. Uncertainty grows exponentially over time; A diffusion pattern produced 30 years from now can change completely with a very small deviation from today's assumptions. That's why long-range scenarios are valuable for a "direction" and a "warning", but not for a binding decision at the parcel level. The fact that AI can easily produce projections to distant years should not make us forget this limit; As the horizon gets longer, it is necessary to write the uncertainty more clearly.
Comparison with indicators
Compare the scenarios not as "good/bad", but with indicators: how many hectares of agricultural land have been opened for construction, how many people have settled in the flood zone, what is the average transportation distance, is there green space per capita, what is the infrastructure cost? AI comes in handy in setting up this set of indicators and comparing scenarios in a tabular form. Which indicator is more important is a public choice; It is the decision of the process, not the AI.
Step by step: AI-powered scenario study
- Define the question and scenarios. For example trend / compact / protection priority.
- List inputs and constraints. Suitability, demand, protected areas.
- Clarify the assumptions for each scenario. Write transparently.
- Run the simulation. (Model with real data; AI in fiction and report.)
- Install a gauge set. Comparison chart with AI.
- Compare scenarios. Show trade-offs, not winners.
- Report ambiguity and boundaries; Leave the decision to the process.
Weak prompt / Strong prompt
Weak prompt: "How will this city grow in 2050?"
AI produces an unsourced, single, misleading “prediction”; There are no assumptions and restrictions.
Strong prompt: "I want to compare three urban growth scenarios: (1) trend (current expansion continues), (2) compact (development is concentrated within the existing city), (3) conservation priority (agriculture and floodplains are strictly excluded). Set up a list of assumptions for each scenario and a comparison table with the following indicators: built-up agricultural area (ha), population settled in the floodplain, average travel distance, green area per capita. Don't pick the winner; explain the trade-offs. This is a scenario, not a prediction." highlight."
The second prompt clarifies the scenarios, assumptions, indicators and trade-off framework.
Four copyable templates
Task: Set up a list of assumptions and indicator comparison tables for the following three growth scenarios. Indicators: [...]. Choosing the winner; Write down the trade-offs of each scenario. Add the note "This is a scenario, not a prediction." Scenarios: [...]
Task: Check the following simulation inputs. Which inputs contain assumptions, which ones are not clear as constraints (area closed to construction), which data up-to-dateness should be questioned? Decision making; just tick.Inputs: [...]
Task: Write a council/stakeholder presentation DRAFT from the scenario comparison results below. Explain the trade-offs of each scenario in plain language. State clearly the uncertainty and that "the decision belongs to the official process." Results: [...]
Task: Explain the terms cellular automaton, growth probability, constraint layer, scenario and indicator in plain Turkish in a dictionary. Emphasize in one sentence why the simulation is a scenario and not a prediction.
A table: scenario types and indicators
Scenario
basic assumption
prominent risk
Indicator to watch
Trend (current)
Propagation continues
Agriculture/nature loss
Built-up agricultural area
compact
Concentration inside
Extreme density/comfort
Green area per person
Protection is priority
Sensitive area is excluded
Housing supply constraint
housing shortage
transportation oriented
Line perimeter improves
Even addiction
Average transportation distance
three mini cases
Case 1 — Scenario thought to be a guess. A municipality produces a single "growth map" and presents it to the council saying "this is the future of the city". One council member asked “what are the assumptions?” he asks; The team realizes that agricultural areas are assumed to be open to construction. When the protection priority scenario is added, the map changes completely. Presenting a single scenario as the future closes the discussion from the beginning.
Case 2 — Making the trade-off visible. One team compares three scenarios with indicators: the conservation scenario saves agriculture but creates a housing deficit; The compact scenario preserves land but increases density. The picture AI sets makes the reality of “there is no free lunch” visible and parliament can make an informed choice.
Case 3 — Flood settlement warning. In a trend scenario, the dashboard reveals that 6,500 people will settle in the flood zone. This number indicates that the constraint layer is loosely defined in the scenario. The team tightens the constraint. The indicator quantifies a risk that will be overlooked.
Common mistakes
- Mistaking the scenario as a guess. The simulation says "may", not "will".
- Presenting a single scenario. Without comparison, preference cannot be discussed.
- Hiding assumptions. The believable map can cover up the hidden assumption.
- Defining constraints loosely. If areas that need to be protected are not clearly excluded, the risk increases.
- Declaring a winner. The indicator shows trade-offs; The choice is the public process.
- Skipping data update. Outdated availability/demand data breaks the entire scenario.
In summary
Urban growth simulation is a powerful way to compare “what if” questions; AI helps in scenario construction, assumption editing, indicator comparison and reporting. But a simulation is a scenario, not a prediction: when the assumptions change, the outcome changes. Run several scenarios instead of a single map, compare against indicators, make trade-offs visible and clearly define constraints (areas to protect). Neither the AI nor the map chooses the winner; The decision is made through the official plan process and expert evaluation.
Application task
Define three growth scenarios for a city/region (e.g. trend, compact, conservation priority). (1) Set up the assumptions of each scenario and an indicator comparison table with the first template. (2) Check the inputs with the second template and capture at least one loosely defined constraint. (3) Produce a parliamentary presentation outline with the third template, emphasizing the “scenario/prediction” distinction. (4) Explain in one sentence which indicator's prominence is a public choice.
checklist
- [ ] I framed the simulation as a scenario, not a prediction.
- [ ] Instead of a single map, I set up several scenarios and compared them.
- [ ] I have clearly written the assumptions of each scenario.
- [ ] I have clearly defined the constraints (areas to be protected).
- [ ] I compared the scenarios with the indicators, I did not declare a winner.
- [ ] I made the tradeoffs visible.
- [ ] I left the decision to the official planning process and expert evaluation.