1. Assistive, not authoritative
GeoAI and GeoML features are designed to assist exploration, analysis, explanation and workflow creation. They do not replace a qualified professional, authoritative source data or a statutory decision-maker.
2. Human review
Users should review AI-generated text, classifications, predictions, summaries and workflow suggestions before relying on them. The person or organization using the output remains responsible for the decision made from it.
3. Data quality and provenance
AI outputs are only as useful as the data, assumptions, prompts and models behind them. Users should consider source date, completeness, spatial resolution, CRS, geographic coverage, sampling bias, missing values and whether the data is appropriate for the question.
4. Explainability and evidence
Where practicable, GeoLayers is designed to retain inspectable source layers, parameters, model diagnostics, workflow steps and map outputs so a conclusion can be reviewed. Users should preserve that evidence when a decision needs to be auditable.
5. Bias and fairness
Spatial data can encode historical inequalities, collection bias and incomplete representation. Predictive or suitability models can amplify those patterns. Users should assess whether criteria or outputs create unjustified disadvantage, especially where decisions affect people, access to services, land, housing or public resources.
6. Personal and sensitive data
Do not provide personal, confidential or sensitive project data to an AI feature unless you are authorized and the feature's data path is appropriate for that information. Organization policies may restrict which projects or fields may be used with AI.
7. High-impact decisions
GeoLayers AI outputs should not be the sole basis for a decision that has a legal or similarly significant effect on an individual where human review or another safeguard is required by law or good professional practice. This includes decisions where data quality, bias or context could materially affect a person's rights or access to important services.
8. Planning, environmental and infrastructure use
AI may help summarize scheme controls, compare scenarios, classify imagery or identify patterns, but users must verify official zoning, title, environmental authorization, engineering capacity, cadastral boundaries and other authoritative requirements before acting.
9. Generative output and intellectual property
AI-generated text or code can resemble existing material or contain errors. Review outputs before publication or deployment and do not assume an AI response automatically gives you rights in third-party material.
10. Security
Do not ask AI tools to expose secrets, API keys, private credentials or unauthorized data. GeoLayers may apply technical boundaries to prevent secret export or unrestricted access to project information.
11. Model and provider changes
AI models and providers may change over time. Results can differ between versions. For reproducible professional work, record the date, relevant model/version information where available, source data and key parameters.
12. Reporting concerns
Report a concerning AI output, suspected privacy issue or security issue to info@geolayers.co.za with enough context to investigate without unnecessarily including sensitive data.