Gains:
- Understand that AI can be wrong about climate zone, invasive status, toxicity and market because it learns from the global average
- Ability to use AI solely as a trait profile and candidate list generator, validating each candidate with local climate, invasive list, toxicity and sapling availability
- Ability to internalize that no unverified plant will be included in the project and that expert and official sources are essential for poisonous/allergenic species.
The most living or dying part of a landscaping project are the plants. An incorrectly chosen tree will dry out in the first dry summer; an invasive ivy overwhelms an entire garden in three years; An allergenic species makes a schoolyard unusable. Plant selection is to determine the plant palette suitable for the climate, soil, function and maintenance capacity of the place. A native species is a plant that grows naturally in a region and belongs to that ecosystem; An invasive species is a plant that comes from outside and spreads uncontrollably, suppressing local species. In this unit, we will discuss why AI is only a candidate list producer in plant selection and why the final decision is made with local verification. This is the most demanding topic of the entire module.
Why ask AI “which plant?” so unreliable?
AI learns from texts around the world; so their recommendations shift to the global average. Yet plant suitability is extremely local: the same species lives in one valley and freezes 30 km away. AI often makes these mistakes:
- Climate zone mixing: Recommends the warm climate plant to the cold region. (Climate/hardiness zone — the lowest temperature range a plant can withstand.)
- Don't assume the invasive species is innocent: You may not know that a species is considered invasive in a particular country/province.
- Lack of knowledge of market reality: The species he recommends may not be available at the local nursery.
- Latin name error: Sometimes produces nonexistent or confused scientific names.
- Soil/water incompatibility: It can recommend plants that require intense water for arid areas.
So the golden rule: AI suggests plant name; you verify it with the local climate zone, invasive species list, soil and seedling market. Unverified plant list does not enter the project.
Caution: Poisonous or allergenic plants (especially in areas where children, patients and elderly people are used) are high risk. It is never enough for the AI to say a plant is “safe”; Confirm toxicity and allergenicity with local/professional source.
The right workflow: from candidate list to validated palette
Step 1 — Clarify the terms. Write down climate, soil, water, sun/shade, function (shade/curtain/color), maintenance capacity and safety constraints.
Step 2 — Ask the AI for a candidate feature profile. Omit the desired characteristics before the species name ("drought tolerant, low maintenance, native, non-allergenic, shade-giving").
Step 3 — Get a candidate list but connect the context. Ask the AI for candidate species suitable for the region; justification for each.
Step 4 — Local verification (most critical step). Confirm each candidate for: (a) local climate zone, (b) national/provincial invasive-species list, (c) native sapling availability, (d) toxicity/allergenicity. Source: university, botanical garden, provincial directorate of agriculture, local nurseryman, relevant official lists.
Step 5 — Maintenance and seasonal plan. Create a plan for watering, pruning and seasonal appearance of the selected palette.
three mini cases
Case 1 — Invader caught. YZ recommended an ornamental vine for a coastal project because it grows quickly. The architect saw from the source of the provincial directorate that the species was on the invasive list in the relevant region and deduced it. It was replaced by a local alternative; After 3 years the garden was stable.
Case 2 — Climate mismatch. A designer wanted to place YZ's proposed warm-climate palm in a continental-cold town square. The senior architect realized that the minimum temperature was below the endurance limit of the species. The palette has been revised from the beginning; There were no deaths in the first winter and the budget was preserved.
Case 3 — Market reality. Of the 12 species suggested by YZ, 4 were not available at all in local nurseries; import costs would exceed the budget by 20%. The team added a constraint to the AI, "only work with strains common in the local market," to refresh the list and keep the cost on target.
Four copyable templates
1) Feature profiling:
Your role: landscape plant consultant. Conditions: [continental climate, -15C in winter, clay soil, sunny, low irrigation, schoolyard -NO allergenic and non-toxic, shade priority]. First, give me the plant FEATURE profiles (not the species name) SUITABLE for this area: which hardiness, which form, which care. Then suggest candidate species that fit these profiles; justification for each.
2) Candidate list + verification alert:
Suggest a list of candidate plants for [region/climate]. For EACH species: why suitable, estimated water needs, possible risk. Add this to the end of the list: remind me that each species must be verified by me for local climate zone, invasive species list, seedling availability, and toxicity.
3) Validation query set:
For this candidate species list, generate control questions that I would ask when validating each species locally: climate zone compatibility, invasive status, toxicity/allergenicity, soil-water compatibility, local sapling availability, and mature size/root spread.
4) Maintenance and season plan:
Create a care and seasonal management plan for the following verified palette [species list]: watering frequency, pruning time, bloom/color periods, and potential problems. State the difference between the first 2 years (holding on) and later.
Weak prompt / Strong prompt
Weak prompt:
Suggest beautiful and hardy plants for my garden.
There is no climate, no land, no security and no market; the output would be foreign to the locale, risky, and possibly unfindable.
Powerful prompt:
Your role: plant advisor. Conditions: Mediterranean semi-arid climate, very hot/dry in summer, light frost in winter, calcareous loam soil, low irrigation budget, public park - no allergenic and poisonous species, priority shade and low maintenance. Candidate lists are made only from species common in this region; Write down water needs and risks for each species. Note that the decision will be verified by me with the invasive list and the local nurseryman.
Climate, land, security, budget and the “local market only” constraint make the output realistic and safe; The verification note puts the responsibility in the right place.
Plant selection risk chart
Subject
Risk of AI errors
verification source
Consequence risk
climate adaptation
high
Climate/hardiness zone data
mass death
invasive status
high
National/provincial invasive list
ecosystem damage
Toxicity/allergy
high
Expert/official source
human health
sapling availability
medium
local nurseryman
Budget/delay
Soil-water compatibility
medium
soil test
poor development
mature size
medium
botanical source
infrastructure conflict
Common mistakes
- Putting the AI list into the project without validating it. This is the most expensive and common mistake.
- Bypassing invasive species control. Fast-growing "practical" suggestions are often invasive.
- Assuming the climate zone. Even within the same country, regions are very different; Check the minimum temperature.
- Overlooking toxicity/allergen. This is a high risk, especially for children and in public areas.
- Forget the market. The kind that cannot be found, even if beautiful, cannot be implemented; Require local availability.
In summary
Plant selection is the most local and risky decision of this profession. Because AI learns from the global average, it can be wrong about climate, invasive status, toxicity, and markets. So use AI only as a feature profile and candidate list generator; Verify each candidate for local climate zone, invasive list, toxicity, and seedling availability. No unverified plants enter the project; Expert and official sources are essential on safety-critical (toxic/allergenic) issues.
Application task
Define the conditions (climate, soil, security, budget) for a site and generate a candidate list with the "Feature profiling" template. Then subject each candidate to validation questions with the “Validation query set” template and remove at least two species from the list (because they are invasive, climate incompatible, unavailable, or toxic); Write the reason.
checklist
- [ ] I used AI only for the candidate list, not for the final decision.
- [ ] I verified each candidate for climate/hardiness zone.
- [ ] I checked the invasive species list.
- [ ] I have confirmed toxicity/allergenicity (especially in public).
- [ ] I checked local sapling availability and mature size.