ResearchTechnology and Digital PracticeAI Tools in Landscape Design Workflows

AI in landscape design practice: uses, limits, checks

What ASLA, regulators and NIST say about AI in design work, plus copyright, confidentiality and a routine for checking AI output before it leaves the office.

Field Method · AI-assisted research roundupPublished Sources checked

Generative AI can draft, summarize, sketch and write code quickly. It cannot hold a license or answer to a client, a board or a court. Every professional statement we could read puts that responsibility on the licensed person. For a practice, the working questions are which tasks the tools may touch, which tools, which data may enter them, and how each output is checked before it leaves the office.

Below are what the professional bodies and regulators say, the risks as NIST describes them, the confidentiality and copyright points, and a verification routine. The routine is our interpretation, built from the sources. This is not legal advice or guidance on professional duty. Questions about your obligations belong with counsel and your licensing authority, which in California is the California Architects Board acting through its Landscape Architects Technical Committee (LATC).

What the professional bodies and regulators say

ASLA. ASLA’s 2025 public policy statement says AI is not a substitute for human judgment, creativity, or ethical and professional responsibility. It says the Professional Landscape Architect must keep responsibility for all deliverables and services that protect public health, safety and welfare, and that it remains the licensed landscape architect’s responsibility to serve in conformance with the standard of care. It names site analysis, climate modeling and construction documentation as areas of opportunity. It warns that unsuitable models or inadequate prompts can limit the efficacy of design solutions, that many models are opaque and grounded in biases, and that data centers use large amounts of electricity and water. ASLA, Artificial Intelligence (2025) ASLA’s news item of December 16, 2025 reports that the Board of Trustees approved the policy at its fall meeting. ASLA news

The regulators’ council. CLARB reports that in January 2026 the leadership of the Interorganizational Council on Regulation approved a joint public statement on AI in the design professions. ICOR’s members are NCARB, CIDQ, CLARB and NCEES. CLARB summarizes the statement as affirming that AI can support practice but not replace professional judgment, ethical responsibility or regulatory accountability. The page does not link to the statement’s text. We did not read the statement itself. CLARB, ICOR meeting highlights

NCARB (architecture). NCARB’s position, first published October 17, 2024 and updated in April 2026, says AI supports but does not replace professional judgment, and that only a licensed practitioner may seal technical submissions. Regulators should not limit the tools practitioners use, and any AI-related regulation must keep the practitioner in responsible control. NCARB says it plans to identify best practices, including whether, when and how AI use and datasets should be disclosed. NCARB position In our reading, that last point means no disclosure standard has been settled at the national licensing level for architecture.

NCEES (engineering and surveying). NCEES Position Statement 6.10, adopted in 2025, is more specific than the others we read. It says licensed professionals keep ultimate responsibility and must not use AI to practice outside their competency. They must perform independent checks of outputs before implementation. Results should be verifiable, with documented methods, data sources and assumptions. Data privacy and security are named as critical. It says member boards should develop specific guidelines or rules. NCEES Manual of Policy and Position Statements, PS 6.10, pp. 46–47 The Idaho engineering board sent the statement to licensees on January 23, 2026 and said it has not been adopted into Idaho law or rule. Idaho board letter NSPE’s position calls for rigorous testing, validation and monitoring of AI systems and for compliance with privacy law. NSPE Position Statement 03-1774

AIA (architecture firms). AIA’s guidance, updated October 22, 2025, is the most practical document we found for a firm. It says AI-generated outputs must be reviewed and validated by qualified professionals and that firms should assign oversight and write policies on approved systems, data privacy, intellectual property and legal standards. It says not to upload confidential client or firm data into unvetted systems and to understand how platforms store, reuse or train on uploaded information. It says to disclose to clients and stakeholders when and how AI is used, and to consider disclosure clauses in consultant agreements. AIA, Guidance for the Responsible Use of AI Its audience is architecture firms. The disclosure advice is AIA’s guidance, not a rule for landscape architects.

California licensing. The LATC’s August 7, 2026 agenda lists a strategic plan item to explore how AI may affect the profession and licensure, and a discussion item asking how the Committee continues to protect public health, safety and welfare in the age of AI tools. The action requested was none. The materials note that the Program Manager would develop an AI reference list for staff. LATC, August 7, 2026 notice and agenda We found no LATC rule or guidance on AI use, and we cannot say what the Committee will adopt.

What the risk framework adds

NIST’s AI Risk Management Framework is voluntary and not specific to any sector. It organizes risk work into four functions: Govern, Map, Measure and Manage. Govern is described as cross-cutting. NIST AI RMF 1.0 NIST’s framework page says AI RMF 1.0 is being revised as part of the White House AI Action Plan. NIST, AI Risk Management Framework Its Generative AI Profile, released July 26, 2024, lists risks that, in our reading, apply to design practice. NIST AI 600-1

  • Confabulation is confidently stated but false output. The profile says it follows from how the models work and is most relevant to open-ended prompts and to domains that need contextual or expert knowledge. It says outputs can include confabulated citations that appear to justify an answer.
  • Human-AI configuration includes automation bias, meaning excessive deference to automated output, which the profile says can worsen confabulation.
  • Data privacy covers leakage and unauthorized use of sensitive data. The profile notes that most developers do not disclose their training sources.
  • Intellectual property risk arises where outputs reproduce training data, and the profile says the status of generated content similar to protected work is being debated in legal fora.

One suggested action is to consider disclosing GAI use to end users, weighing purpose, context, risk and audience. The profile says it was developed under Executive Order 14110. Executive Order 14148 of January 20, 2025 revoked that order. Executive Order 14148, 90 FR 8237 NIST’s framework page still lists the July 2024 profile and shows no revised version of it.

Where AI can help, and what to check

The sources support using AI for first-pass work with a qualified person reviewing it. ASLA names site analysis, climate modeling and documentation. AIA’s examples include concept generation, faster visualization, specification drafting and code summaries. None of the sources measures how accurate these tools are on landscape tasks, so the table below is our judgment, not a finding.

Task What we would use it for What we would check
Research Finding leads, summarizing long documents Open every cited source and confirm the number or passage is there
Option generation Prompting alternatives for a program or layout Program, grades, access, plant suitability, local standards
Option images Mood and massing images for early discussion Label as concept art. Check for impossible geometry and invented plants or products
Documentation Draft notes, schedules, meeting summaries, specifications Compare against the adopted edition of each code and standard, and your own drawings
Code Scripts for takeoffs, data cleanup, batch tasks Run on a case with a known answer. Read the code. Keep it in version control

We found no primary source that addresses AI for design code or scripts specifically, so the last row applies the general verification principles above.

Confidentiality and vendor terms

In our interpretation, a prompt sent to a hosted tool should be treated as sharing that information with the vendor. NCEES names privacy and security as critical, and AIA says not to put confidential client or firm data into unvetted systems. The FTC’s staff post of January 9, 2024 says customers using hosted models may expose sensitive information such as internal documents, that vendors who break privacy commitments can face liability, and that “There is no AI exemption from the laws on the books.” FTC, AI Companies: Uphold Your Privacy and Confidentiality Commitments The post is addressed to AI vendors. It does not set a rule for design firms.

Vendor terms differ by product and plan. We read two pages, and they cover different tiers: Anthropic’s page covers its commercial products, and Google’s covers Gemini Apps used with personal accounts. They do not compare like with like, and we do not rank them. Field Method drafts its articles with Claude, from Anthropic. We say so here so readers can weigh our choice of examples.

  • Anthropic’s privacy page, dated August 18, 2026, covers commercial products such as Claude for Work and the API. It says Anthropic does not train on their inputs or outputs by default, but feedback submitted or data a customer opts into sharing can be used, and feedback conversations can be stored for up to five years. Consumer plans are covered in a separate article we did not read. Anthropic Privacy Center
  • Google’s Gemini Apps Privacy Hub, last updated October 26, 2026 as displayed, covers personal Google Accounts. It says work or school accounts may be subject to different data handling terms, set out in a separate Workspace privacy hub we did not read. It says that when the Keep Activity setting is on, some chats are reviewed by human reviewers to improve Google services, that reviewed chats are kept up to three years, and it advises against entering confidential information you would not want a reviewer to see. Google, Gemini Apps Privacy Hub

Terms change. Before client or project data enters any tool, read the current terms for the exact plan, check your client contract and nondisclosure agreements, and ask counsel where they conflict. Public data, such as USGS or county layers, carries a different risk than a client’s survey or fee proposal. That distinction is our interpretation.

The U.S. Copyright Office’s January 2025 report concludes that copyright does not extend to purely AI-generated material or material where human control over the expressive elements is insufficient. It also concludes that using AI to assist rather than stand in for human creativity does not affect protection, that human authors are entitled to copyright in the creative selection, coordination or arrangement of material in outputs and in creative modifications of them, and that prompts alone do not currently provide sufficient control. Each case is analyzed individually. Copyright Office, Part 2: Copyrightability

For registration, the Office’s March 16, 2023 guidance says applicants have a duty to disclose AI-generated content and briefly explain the human author’s contributions. AI-generated content that is more than de minimis should be excluded from the claim, and an AI system should not be listed as an author. Copyright Office, Registration Guidance As shown on November 14, 2026, the Office’s AI page lists no later registration guidance. It lists Part 3, on training, as a May 9, 2025 pre-publication version, with no substantive changes expected in the final version. Copyright Office, Copyright and AI

In our interpretation, a rendering produced from a prompt alone may carry little protectable authorship, while the drawings, selections and edits a designer makes keep theirs. Keep records of what the designer did. Rules outside the United States differ. Ownership of AI output is also set by vendor terms and your client agreement, so have counsel review both.

A verification routine

This is our interpretation, drawing on NCEES section C and on the NIST profile. It is not a standard.

  1. Record the use. Note the tool, plan, date, purpose, inputs and reviewer, so methods, data sources and assumptions can be documented.
  2. Verify facts at the source. Open each citation. A citation that looks right can be invented.
  3. Recompute numbers. Redo quantities, areas and calculations by an independent route.
  4. Read the adopted code. Use the edition your authority having jurisdiction has adopted, not a summary from a model.
  5. Check against the site. Compare statements about conditions with survey data, source layers and a site visit. An image has no datum and no scale.
  6. Use a reviewer with the competence. NCEES says licensed professionals must not use AI to practice outside their competency.
  7. Label and disclose. Mark concept images as AI-assisted. Follow your client contract and consider AIA’s advice to tell clients when and how AI is used.
  8. Seal or stamp only after review. The statements we read keep responsibility with the licensed professional, and NCARB says only a licensed practitioner may seal technical submissions. In our reading, that holds whatever tool helped. Your board decides what a seal or stamp requires.

How Field Method uses AI

Field Method drafts these research articles with AI (Claude, from Anthropic). Each draft then goes through an independent AI source check, which re-opens every cited source, and an independent check from a practitioner’s point of view, followed by a fixing pass. Articles are published without a person approving each one beforehand. Items that need human judgment are recorded and reviewed after publication, and articles are revised when reviews find problems. None of this makes an article correct. Check the sources yourself before relying on any figure here.

What we could not settle

  • The text of the ICOR statement.
  • Whether any California board or the LATC will adopt AI rules, and whether any state requires disclosure to clients.
  • Vendor terms beyond the two pages above, including Anthropic’s consumer plans and Google Workspace. Two other vendor pages could not be opened.
  • What professional liability insurers require or exclude for AI-assisted work. Ask your insurer or broker.
  • CLARB’s gated ethics resource for record holders.
  • Whether the Copyright Office has updated registration guidance or its Compendium since 2023, and the final version of Part 3 on training.
  • Measured accuracy of AI tools on landscape tasks, which no source we read provides.
  • What the revised AI RMF will contain, and whether the generative AI profile will be revised or replaced.

Using this in Field Method

Site Dossier assembles public layers for a San Diego County point, with each layer’s source and date in an evidence register. That register gives a reviewer something concrete to check an AI-written site summary against. It is a mapped starting record for pre-design, not a survey, and it covers San Diego County only.

Sources

  1. ASLA — Artificial Intelligence (2025), public policy statement
  2. ASLA — ASLA Adopts New Public Policy on Artificial Intelligence
  3. CLARB — Highlights from the January 2026 ICOR Leadership Meeting
  4. NCARB — Position on the Use of Artificial Intelligence in the Architectural Profession
  5. NCEES — Manual of Policy and Position Statements (revised August 2026), PS 6.10
  6. Idaho Board of Professional Engineers and Professional Land Surveyors — Letter enclosing NCEES Position Statement 6.10, Responsible Use of Artificial Intelligence in Engineering and Surveying
  7. NSPE — Position Statement No. 03-1774, Artificial Intelligence
  8. AIA — Guidance for the Responsible Use of AI by Architecture and Design Firms
  9. California Architects Board, Landscape Architects Technical Committee — Notice and Agenda, August 7, 2026
  10. NIST — Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1
  11. NIST — AI Risk Management Framework
  12. NIST — Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1
  13. Executive Order 14148, Initial Rescissions of Harmful Executive Orders and Actions (90 FR 8237)
  14. Federal Trade Commission — AI Companies: Uphold Your Privacy and Confidentiality Commitments
  15. Anthropic — Is my data used for model training? (Privacy Center)
  16. Google — Gemini Apps Privacy Hub
  17. U.S. Copyright Office — Copyright and Artificial Intelligence, Part 2: Copyrightability
  18. U.S. Copyright Office — Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence
  19. U.S. Copyright Office — Copyright and Artificial Intelligence

These tools support site review and design studies. Read each tool’s stated capabilities and limits before use.

Explore Site Dossier
Back to Research