Gains:
- Ability to define the role of AI in data preparation and scenario generation in population projection, density distribution and infrastructure demand forecasting
- Ability to clearly track assumptions, uncertainty range, and method limits in forecast outputs
- Ability to understand that official population and statistical data (e.g. TÜİK) is binding on the AI estimate and that AI does not produce numbers.
Every plan starts with a question: how many people will live here in the future, where will they be concentrated, how many schools, water, roads, green spaces will they need? The answers to these questions are population projection (estimating the future population), density analysis (how the population is distributed in space) and demand forecast (infrastructure and equipment needs according to the population). AI is a powerful aid in data curation, scenario generation and reporting in these jobs. But there is a critical limit: AI does not produce reliable numbers. The binding source of population data is official statistics (TUIK and address-based population registration system in Türkiye). AI can create scenarios with assumptions based on this data, but a population figure it makes up cannot be put into the plan. This unit teaches you where to safely use AI in the forecasting process and where to turn to official data.
Logic of population projection
There are several classical methods of population estimation and they all rely on assumptions:
- Arithmetic/geometric increase: Extends the past growth rate into the future. Simple but misleading in rapidly changing regions.
- Component method (cohort-component): Models birth, death and migration separately. It is more realistic but requires more data.
- Trend/ratio methods: Uses the share of the region within the province/country.
Each method relies on a set of assumptions: will immigration continue, will fertility decline, will industrial investment come? Very different futures emerge from the same past data, with different assumptions. So there is no single "correct number"; There is a range and scenarios. The most valuable contribution of AI is to regularly set up these scenarios, compare them and write down the assumptions transparently.
Attention: It is a serious mistake to ask AI "what will be the population of this district in 2040" and put the single number in the plan. The AI may have hallucinated that number. The correct way is to provide official historical data, generate a range of scenarios with clear assumptions, and calculate the final figure with a verified method.
A particularly dangerous situation in planning is the over-projection of population assumptions. If a plan is prepared with a population target that is too high, it will produce too many development rights, too much infrastructure, and vacant investments; This is both a waste of public resources and a ground for speculation. Conversely, underestimation leaves infrastructure inadequate. Because AI can easily generate multiple scenarios, it helps make these two risks visible: juxtaposing the infrastructure and development rights outcome of each scenario and asking “what do we lose if this assumption doesn't come true?” You can ask the question. Population forecasting is therefore not merely a matter of technique but also of resources and justice.
Density and demand
Population is not a single number, but a pattern distributed over space. Gross density (person/hectare, over the entire area) and net density (over the residential area only) are different things, and if mixed, the plan collapses. Demand forecasting depends on density: the more people there are in a neighborhood, the more school classrooms, the more water, the more parking, the more green space is needed. In these calculations, standards (such as m² of green space per person, classrooms per student) come from the legislation in force; AI can remember the standard, but its value must be confirmed from the official text.
Step by step: Prediction process with AI
- Collect official data. TURKSTAT historical population, age structure, migration data.
- Choose method. Method(s) appropriate to the dynamics of the region.
- Clarify assumptions. Growth rate, migration, fertility — write clearly.
- Have AI create a scenario. Low/medium/high scenario, assuming each.
- Show range and uncertainty. Not a single number, but a band.
- Turn it into demand. Infrastructure/equipment requirement with density and standards.
- Verify. Recalculate the figure by method; Confirm standards from legislation.
Weak prompt / Strong prompt
Weak prompt: "Predict the future population of this district."
AI makes up an unsourced number; Which method, which assumption is unclear.
Strong prompt: "Below is the official annual population data of [district] for 2000-2023 (source: TÜİK). Set up three scenarios: low (immigration slows down), medium (current trend continues), high (migration increases with new industrial investment). Write the assumption clearly for each scenario, use the geometric increase method and produce a 5-year table until 2040. State the uncertainty of each scenario. Don't add data from your own head; just use the series I gave. Data: [...]"
The second prompt method clarifies the assumption, scenario structure, and data limit; The output becomes auditable.
Four copyable templates
Task: Generate three scenarios (low/medium/high) from the official historical population series below. For each scenario: assumption, method used, 5-year projection table, uncertainty note. Just use the data I gave you; don't fabricate data. Series/source: [...]
Task: Check the assumptions, method, and uncertainty statement in the population estimate report below. Flag missing or hidden assumptions. He warns that places presented as "single exact number" should be "a range".Report: [...]
Task: DRAFT the infrastructure/equipment request from the population and density data below (school classrooms, green areas, parking lots, etc.). Give the standard used in each account with the note "must be confirmed by legislation". Maintain the gross/net density distinction. Data: [...]
Task: Explain the terms gross density, net density, population projection and component method in plain Turkish in a glossary table. Emphasize the difference between them in one sentence. Don't attribute numbers.
A table: method, data and limit
Method
Data required
Where it fits
limit
Arithmetic increase
Less (two dates)
stationary zone
Rapid change is misleading
geometric increase
past series
Steady growth
Assumes infinite growth
Component (cohort)
Age, birth, death, migration
detailed planning
Requires a lot of data and assumptions
Ratio/share
City-county series
regional planning
dependent on upper scale estimation
three mini cases
Case 1 — Hallucinated population. A planner directly asks the AI “2045 population”; The AI confidently says "324,000". When the planner makes a geometric projection with TÜİK data, the range of 268,000-341,000 comes out with the same assumptions. AI's odd number was within this range, but coincidentally; On another question, he might have given a completely outlandish value. Lesson: the number is generated by the method, not requested from the AI.
Case 2 — Gross/net confusion. A team sets a target of "150 people/hectare" for a region but does not specify whether it is gross or net. Confusion grows in the report they print to the AI; When the value considered as net density is taken as gross, the need for housing space doubles. Clarifying the term from the beginning would have prevented the crisis.
Case 3 — Scenario transparency. A municipality presents a three-scenario population study to the council; The assumption and range of each scenario are clear. The parliament can discuss "what will happen to the school deficit if the high scenario occurs?" Presenting a scenario rather than a single number makes the plan flexible and defensible.
Common mistakes
- Putting the odd number coming out of the AI into the plan. The population figure is produced using official data and method; Not required of AI.
- Hiding the assumption. Every projection is based on an assumption; If it is hidden, the result is misleading.
- Mixing gross/net density. The classic mistake that spoils the entire demand calculation.
- Reducing uncertainty to a single number. The future is an interval; tape must be presented.
- Not confirming the standard. Values such as green area per capita should be verified by legislation.
- Ignoring migration and dynamics. The simple method is misleading in the fast growing/shrinking region.
In summary
Population, density and demand forecasting are the basis of every plan and are all based on assumptions. AI is a powerful aid in this process in organizing data, building scenarios, writing assumptions transparently and reporting. But AI doesn't produce reliable numbers; The binding source of the population is official statistics (TURKSTAT). The right way is to provide official historical data, produce a scenario range with clear assumptions, separate gross/net density and confirm the standards from the legislation. Present hypothetical scenarios and a range of uncertainty, not a single "correct number."
Application task
Take a historical population series for a region (actual or sample). (1) Have the AI build three scenarios with the initial template; Check each assumption. (2) Check the report's uncertainty and assumption statements with the second template and correct a "single absolute number" error. (3) Draft a claim with the third template and check that you have allocated the gross/net density correctly. (4) List which standards and which legislation you will confirm.
checklist
- [ ] I received the population data from official sources (TUIK, etc.).
- [ ] I calculated the final figure using the verified method, I did not ask it from the AI.
- [ ] I have clearly written the assumption of each scenario.
- [ ] I presented the result as an uncertainty range, not a single number.
- [ ] I separated gross and net density.
- [ ] I confirmed the standards in demand calculations from the legislation.
- [ ] I took migration and regional dynamics into account in choosing the method.