Aipedia: Landscape architects

AI in landscape character and sensitivity studies

Aipedia, by the Apex broking team · Last reviewed 2026-10-30

AI can speed up mapping and drafting for landscape character assessments and sensitivity studies, but the classification and judgements must be verifiable and the practitioner's own. Councils rely on these studies as local plan evidence, so a weak one can fail at examination.

Part of: AI in Landscape architects

In short

Landscape character assessments and sensitivity or capacity studies may be prepared with AI help: automated classification of land cover and topography, clustering into character types and drafted area descriptions. The risks are generic text that could describe anywhere, data sources that cannot be traced, and scoring nobody can explain. Under the LI Code, Rule 6 requires competence, R11g(a) transparency and Rule 10 impartiality, while R4g(c) agreements should set scope and limitations. If a study fails at examination, the commissioning council or developer may claim the cost of redoing it, which PI usually addresses subject to the wording.

How AI is used in character and sensitivity work

Character assessments start with layers of data: geology, landform, land cover, settlement pattern, field boundaries, designations and historic landscape characterisation. AI tools can cluster those layers into draft character types, suggest area boundaries from imagery and draft descriptions of key characteristics, perceptual qualities and forces for change. For sensitivity and capacity studies supporting site allocations, practices use AI to score parcels against criteria and to generate the narrative that justifies each score.

The speed is attractive on district-wide commissions with tight budgets. The risk is that the outputs look complete before anyone has walked the ground.

The duties that apply

Under the LI Code of Practice, four obligations shape how AI can be used on these commissions:

Rule 6: only provide services you are competent to deliver, which includes understanding how an AI classification reached its result;
R11g(a): honesty and transparency in all aspects of the work, including the methods and data behind the study;
Rule 10 and R10g(a): impartiality, which matters when a developer commissions a sensitivity study for land it wants allocated;
R4g(c) and R4g(f): a written agreement setting scope and limitations, and confidentiality for client data and licensed datasets put into AI tools.

If the study later becomes appeal or NSIP evidence, the Planning Inspectorate's AI casework guidance requires intentional AI use in drafting or analysis to be declared.

What can go wrong

Three weaknesses recur. First, generic descriptions: AI-drafted text tends to use the same phrases for every area ("gently rolling farmland with a strong sense of place"), which undermines the distinctions the study exists to draw. Second, unverifiable data sources: a classification built from a tool's internal model or an unknown dataset cannot be replicated by the council or challenged by objectors. Third, scoring without reasoning: if parcels are rated by a model, the practitioner may be unable to explain under questioning why one parcel is high sensitivity and its neighbour moderate.

Illustrative scenario. A practice prepares a settlement-edge sensitivity study for a district council using AI clustering and drafted descriptions. At examination, objectors show that several parcels were scored on outdated land cover and that two character areas share near-identical text. The inspector finds the evidence not robust and the council must commission a new study, delaying the plan. The council then seeks its additional costs from the practice.

Related evidence issues are covered in AI in LVIA chapters and, for habitat data, in AI habitat mapping from satellite and drone data.

How PI usually responds

A council or developer claim for the cost of a replacement study, or for losses from a delayed allocation, is a professional negligence claim that landscape PI will usually consider, subject to the terms. Questions that commonly arise include whether the loss flows from the study or from wider plan issues, whether the appointment capped liability, and whether the policy limit matches the size of the commission. Public sector frameworks often set their own PI requirements, even though the LI sets no minimum. See PI for landscape architects in 2026 and does landscape architects' PI cover AI errors?.

Practical controls

Studies that will be tested at examination benefit from these controls:

list every dataset used, with source, date and licence, so the method can be replicated;
treat AI boundaries and clusters as a draft to be checked by fieldwork, not a result;
write criteria and scoring definitions before any parcel is scored;
edit AI descriptions until each area reads as distinct and site-specific;
record in the report how AI was used and what the practitioner verified;
agree in the appointment who owns the data and outputs and what the limitations are.

How Apex helps

Apex arranges PI for landscape planners and practices that prepare evidence for councils and developers. A broker can check your limits against framework requirements and how your policy treats AI-assisted studies. See also generative AI in masterplanning, AI urban greening scores and the AI in landscape architecture overview.

Sources

Frequently asked

Can AI produce a landscape character assessment?

AI can help classify data and draft descriptions, but the assessment still needs fieldwork, professional judgement and a replicable method. Under LI Rule 6 the practitioner must be competent to deliver the study, and R11g(a) expects transparency about how it was produced.

Why do AI-assisted studies fail at examination?

Commonly because the descriptions are generic, the data sources cannot be traced or replicated, or the practitioner cannot explain how parcels were scored. Inspectors and objectors test whether evidence is robust, and AI outputs that have not been checked against the ground are an easy target.

Does impartiality matter for developer-commissioned studies?

Yes. LI Rule 10 and R10g(a) require impartiality, and a sensitivity study prepared for land a client wants allocated must still reflect the practitioner's own judgement. AI drafting can drift towards the outcome the prompt suggests, so the reasoning should be reviewed critically before issue.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.