Environmental Impact Statement and Decision Support

Introduction

An Environmental Impact Statement (EIS) is the principal document through which a proposed construction project is described, its reasonably foreseeable environmental effects are analysed, alternatives are compared, and mitigation and monitoring commitments are presented for regulatory and public review. It is more than a technical report. A good EIS is an evidence-based decision instrument that connects project design, environmental science, stakeholder concerns, legal requirements, risk management, and approval conditions.

For construction professionals, the EIS is important because decisions made during planning influence impacts that may persist throughout construction, operation, maintenance, and decommissioning. Poorly defined project boundaries, weak baseline data, unsupported predictions, or vague mitigation commitments can result in regulatory delay, litigation, cost escalation, reputational damage, and environmental harm. Conversely, a clear EIS can improve project design and demonstrate that environmental risks have been responsibly managed.

This chapter focuses on preparing and using an EIS, communicating uncertainty, integrating technical findings, supporting regulatory review, and applying transparent decision tools to approval conditions and project selection. Exact statutory requirements differ among jurisdictions; therefore, practitioners must always consult the applicable environmental legislation, regulations, permits, and agency guidance.

Learning objectives

By the end of this chapter, students should be able to:

1. Explain the purpose, structure, and legal-administrative role of an EIS.
2. Distinguish project effects, impacts, risks, mitigation, residual effects, and cumulative effects.
3. Integrate findings from disciplines such as ecology, hydrology, air quality, noise, traffic, heritage, and socioeconomics.
4. Communicate uncertainty using assumptions, confidence levels, sensitivity analysis, and adaptive management.
5. Compare alternatives using transparent multi-criteria and risk-informed decision methods.
6. Formulate measurable approval conditions, monitoring requirements, and compliance triggers.
7. Evaluate the quality, ethics, limitations, and governance of an EIS.

1. The EIS as a decision document

Environmental impact assessment (EIA) is a process. The EIS is the principal evidence record produced within that process. It normally supports a decision by a competent authority, but it does not itself grant approval. Depending on the jurisdiction, the decision may be made by an environmental regulator, planning authority, minister, public inquiry, or another legally designated body.

A credible EIS answers five connected questions:

• What is proposed, where, when, and at what scale?
• What environmental and social conditions currently exist?
• What changes could the project cause, directly or indirectly?
• How can adverse effects be avoided, minimized, restored, compensated, or monitored?
• Should the project proceed, proceed with conditions, be redesigned, or be rejected?

The EIS should be traceable. A reviewer should be able to follow a chain from a project activity, through a pathway and receptor, to a predicted effect, its significance, proposed mitigation, residual effect, monitoring indicator, and enforceable condition.

1.1 Typical EIS structure

A jurisdiction may prescribe a particular format, but a comprehensive EIS commonly includes:

The executive summary must not be a marketing statement. It should disclose significant adverse effects and unresolved uncertainty as clearly as benefits.

2. Defining the assessment before collecting data

2.1 Project definition and boundaries

The project description must include not only the main structure but also associated and enabling works: access roads, borrow pits, quarries, worker accommodation, transmission lines, pipelines, temporary laydown areas, spoil disposal, water abstraction, traffic routes, and restoration activities. Omitting associated works can produce an artificially low impact estimate.

Four boundaries should be specified:

• Spatial boundary: the footprint, surrounding influence area, watershed, airshed, ecological system, transport network, or cultural landscape affected.
• Temporal boundary: construction, operation, maintenance, accidents, closure, and post-closure conditions.
• Administrative boundary: jurisdictions, landowners, communities, and agencies with authority or interests.
• Analytical boundary: the issues, receptors, pathways, scenarios, and alternatives included in the assessment.

Boundaries should follow environmental processes rather than convenient property lines. For example, a river impact may require a catchment-scale assessment, while a dust assessment may require modelling beyond the site boundary.

2.2 Scoping and significance questions

Scoping identifies valued environmental and social components (VECs), impact pathways, data needs, and matters that do not require detailed analysis. A useful screening question is:

Could a project activity plausibly alter the condition, availability, safety, use, cultural value, or resilience of a valued receptor?

Significance is not identical to magnitude. A small change in a highly sensitive wetland may be more important than a larger change in a degraded industrial area. Significance commonly reflects magnitude, extent, duration, reversibility, probability, sensitivity, legal protection, distribution of benefits and burdens, and stakeholder or cultural importance.

A qualitative formulation can be expressed as:

where S is significance; M is magnitude; E is spatial extent; D is duration; R is reversibility; P is probability; and V represents receptor vulnerability, value, or sensitivity. This is a conceptual relationship, not a universal numerical law. The variables should not be multiplied or assigned scores unless the assessment method transparently defines how that is done.

3. Baseline evidence and impact prediction

Baseline information establishes the reference condition against which change is evaluated. It should describe variability, not merely a single favourable measurement. Seasonal surveys, historical records, remote sensing, community knowledge, laboratory analysis, and field observations may all be relevant.

Baseline quality depends on:

• appropriate sampling locations and times;
• validated instruments and laboratory methods;
• documented detection limits and missing data;
• representative meteorological, hydrological, ecological, and social conditions;
• quality assurance and quality control;
• identification of existing pressures and trends; and
• clear separation of measured facts, assumptions, and professional judgement.

A baseline is not necessarily a pristine condition. Existing pollution, climate change, traffic, invasive species, or prior development may already influence the receptor. The EIS should distinguish the current baseline from a no-project future baseline, because environmental conditions may change even if the project is not built.

3.1 Pathway–receptor analysis

Impact prediction should identify:

1. Source or activity: excavation, blasting, discharge, clearing, traffic, pumping, or energy use.
2. Pathway: air, surface water, groundwater, soil, noise transmission, visual exposure, traffic, or social interaction.
3. Receptor: people, habitats, species, water users, cultural resources, infrastructure, or ecosystem services.
4. Change: concentration, flow, noise level, habitat area, risk, access, employment, or cultural use.
5. Consequence: ecological, health, social, economic, cultural, or regulatory outcome.

This structure prevents unsupported statements such as “the project will have minimal impact.” It requires the author to show how an effect could occur and why the predicted magnitude is credible.

3.2 Cumulative and induced effects

Cumulative effects arise when the project combines with other activities or environmental trends. Examples include several developments drawing water from one aquifer, multiple roads fragmenting habitat, or construction emissions combining with existing urban pollution. Induced growth, secondary development, and displacement of activities should be considered where reasonably foreseeable.

Cumulative assessment should identify other projects, shared receptors, temporal overlap, additive or interactive mechanisms, and the capacity of the environmental system to absorb further change. It is often more uncertain than project-specific assessment, so assumptions and scenario boundaries must be explicit.

4. Mitigation hierarchy and residual effects

The preferred mitigation sequence is generally:

1. Avoid the impact through site selection, timing, design, or cancellation.
2. Minimize the impact through engineering and operational controls.
3. Restore or rehabilitate affected environments.
4. Offset or compensate, where legally permitted and ecologically credible, for residual effects that cannot reasonably be avoided or restored.

Mitigation must be specific. “Implement dust control” is weak. A stronger commitment identifies water application or enclosure methods, meteorological limits, responsible personnel, inspection frequency, performance criteria, and corrective action.

Residual effects are those remaining after mitigation. An EIS should not classify an impact as insignificant merely because mitigation is proposed; it should assess whether the mitigation is technically feasible, funded, enforceable, proven, and likely to work under the stated conditions.

A commitment register is useful:

5. Communicating uncertainty honestly

Uncertainty is unavoidable because baseline data are incomplete, environmental systems vary, models simplify reality, and future behaviour or climate conditions cannot be known exactly. Uncertainty is not evidence that no conclusion can be reached. It should be characterized and managed.

5.1 Types of uncertainty

• Measurement uncertainty: limitations of instruments, laboratories, sampling, and detection.
• Natural variability: seasonal, spatial, or interannual variation.
• Model uncertainty: simplifications, parameter choices, and structural assumptions.
• Scenario uncertainty: unknown future projects, climate, demand, technology, or operations.
• Knowledge uncertainty: insufficient understanding of ecological or social relationships.
• Decision uncertainty: disagreement about values, acceptable risk, or distributional fairness.

The EIS should state the direction and possible consequence of uncertainty. For example, underestimating groundwater recharge may make a predicted drawdown appear smaller than it could be; overestimating attenuation may understate downstream exposure.

5.2 Sensitivity and scenarios

Sensitivity analysis tests whether a conclusion changes when an important assumption changes. If a model predicts concentration C from emission rate Q and dispersion factor k, a simple relationship might be:

Here C and Q must use compatible units, while k has units that convert emission rate to concentration. The relationship is illustrative and should only be used when the underlying model supports proportionality. If k varies with wind, distance, terrain, or atmospheric stability, those dependencies must be represented rather than hidden in one constant.

A sensitivity test can compare a base case with low-recharge, high-demand, or extreme-weather cases. Results should be reported as ranges or scenarios where appropriate, not as false-precision point estimates.

5.3 Confidence statements

Confidence should relate to evidence and model performance. A useful statement identifies the conclusion, evidence, limitation, and management response:

Confidence is moderate that average construction noise will meet the criterion at the nearest residence, based on validated modelling and two representative surveys; confidence is lower during simultaneous piling and adverse wind conditions. Real-time monitoring and a stop-work trigger are therefore required during piling.

Adaptive management is appropriate when uncertainty is material but manageable. It requires pre-defined indicators, thresholds, responsibilities, decision dates, and actions. It is not a substitute for adequate assessment or an excuse to defer all decisions.

6. Integrating multidisciplinary findings

An EIS is often prepared by specialists, but the decision concerns the project as a whole. Integration requires a common project description, shared assumptions, compatible spatial and temporal boundaries, and explicit treatment of interactions.

For example, a construction camp may increase water demand, wastewater generation, traffic, waste, noise, and pressure on local services. Each specialist report may find a manageable effect, while their combined effect may exceed local capacity. Integration should therefore include:

• a common impact register;
• GIS layers and consistent receptor locations;
• cross-disciplinary review workshops;
• a shared assumptions and data register;
• interaction and cumulative-effect analysis;
• a consolidated mitigation schedule; and
• a plain-language synthesis for decision-makers.

Conflicts between studies should be disclosed and resolved through documented evidence, not concealed by selecting the most convenient result. The lead author is responsible for coherence, but specialists remain responsible for the validity of their technical conclusions.

7. Alternatives analysis

Alternatives are central to decision quality. A reasonable alternatives analysis may include different sites, layouts, technologies, construction methods, scales, schedules, access routes, and the no-project alternative. The no-project case should not be treated as automatically environmentally neutral; it may involve continued existing problems or foregone benefits.

Alternatives should be compared using the same baseline, scenarios, criteria, and level of detail. The analysis should explain why alternatives were screened out and whether design changes emerged from the comparison.

7.1 Multi-criteria decision analysis

Multi-criteria decision analysis (MCDA) can structure comparisons when effects cannot be reduced to money. Suppose alternatives are scored from 0 to 100 against normalized criteria. A simple weighted sum is:

where T_j is the total score for alternative j; w_i is the weight of criterion i; s_ij is the score of alternative j on criterion i; and n is the number of criteria. The weights should satisfy 0 ≤ w_i ≤ 1 and ∑w_i = 1. Scores must be defined so that a higher score consistently means better performance.

The method assumes that criteria are sufficiently independent for additive scoring and that the scales and weights are defensible. It can create misleading precision if scores are arbitrary, distributional impacts are hidden, or legal prohibitions are treated as tradeable preferences. A legally prohibited alternative should normally be screened out rather than compensated by high economic scores.

#### Worked example: comparing two construction sites

Assume two legally feasible sites are scored as follows:

For Site A:

For Site B:

Site A ranks higher under these assumptions. However, the result is not a scientific truth. A sensitivity test should vary weights and scores. If a small change reverses the ranking, the decision should disclose that instability and may require further investigation or a precautionary choice.

8. Risk-informed decisions and approval conditions

Risk assessment distinguishes the possibility of an event from its consequence. A common qualitative matrix classifies risk using likelihood and consequence categories. Such matrices are communication tools, not universal laws; category definitions must be documented.

For quantitative expected-loss analysis, where credible probabilities and consequences are available:

where E(L) is expected loss; p_k is the probability of scenario k; and L_k is the consequence or loss for that scenario. The probabilities should cover the defined scenario set and be appropriate to the time period. Expected loss can obscure rare catastrophic outcomes, irreversible ecological damage, and unequal distribution, so it must not replace legal standards, precaution, or stakeholder consideration.

Approval conditions should be:

• legally authorized and clearly worded;
• specific, measurable, achievable, relevant, and time-bound;
• linked to an identified impact or uncertainty;
• assigned to a responsible party;
• supported by monitoring and reporting;
• enforceable through inspection or evidence; and
• accompanied by consequences or corrective action for non-compliance.

Conditions can address design, construction methods, timing windows, emission limits, ecological buffers, worker and community safety, grievance processes, financial assurance, restoration, monitoring, reporting, independent auditing, and adaptive responses. A condition should not merely repeat an aspiration from the EIS.

9. Regulatory review and public participation

Regulatory review tests completeness, legal adequacy, scientific credibility, alternatives, mitigation, consultation, and consistency with plans and standards. Regulators may request additional information, require revisions, impose conditions, refer the proposal to another authority, or reject it.

An efficient review process benefits from:

• a compliance matrix mapping each requirement to the EIS section and evidence;
• a glossary and consistent terminology;
• a data and modelling audit trail;
• clear separation of fact, prediction, judgement, and commitment;
• response-to-comments tables;
• accessible non-technical summaries; and
• version control for drawings, assumptions, and mitigation measures.

Public participation is an evidence and legitimacy function, not a public-relations exercise. Local and Indigenous knowledge may identify seasonal use, hazards, cultural values, or ecological changes not captured by short surveys. Engagement should be early, accessible, culturally appropriate, and responsive. Consultation does not necessarily mean agreement, but the EIS should show how concerns influenced design or why they could not be accommodated.

10. Ethics, governance, and professional responsibility

EIS authors have a duty of competence, honesty, independence, and traceability. Ethical failures include selective data reporting, understating uncertainty, presenting mitigation as certain without evidence, excluding affected groups, or allowing commercial pressure to determine technical conclusions.

Good governance includes:

• declaring conflicts of interest;
• documenting data ownership and limitations;
• protecting confidential or culturally sensitive information;
• avoiding discriminatory distribution of environmental burdens;
• ensuring accessibility for affected communities;
• maintaining records that support audit and appeal; and
• assigning accountability for commitments after approval.

Environmental justice requires attention to who receives project benefits, who bears risks, and whether vulnerable groups have meaningful participation. A project with a positive average outcome may still be unacceptable if severe burdens are concentrated on a group with limited power.

11. Technology and contemporary developments

GIS supports spatial overlay, receptor mapping, alternatives analysis, and cumulative assessment. Remote sensing and drones can improve land-cover monitoring, but their resolution, classification accuracy, weather limitations, and permissions must be reported. Continuous sensors can provide near-real-time evidence for dust, noise, water quality, or vibration, provided calibration, data security, maintenance, and quality control are addressed.

Artificial intelligence can assist with document search, image classification, anomaly detection, scenario exploration, and public-comment coding. It should not replace professional judgement or be treated as an impartial decision-maker. AI-generated outputs require validation, documentation of training data and uncertainty, protection of personal and culturally sensitive information, and human review. Automated ranking can reproduce bias if historical data reflect unequal participation or enforcement.

Climate change should be incorporated into both impact prediction and project resilience. Baselines may be non-stationary, and extreme rainfall, heat, drought, wildfire, sea-level rise, or changing species distributions may alter impact pathways. The EIS should explain the climate scenarios used and distinguish project effects on climate from climate effects on the project.

12. Illustrative case study: proposed urban flood-control corridor

Background

A municipality proposes a 4-kilometre flood-control corridor along a developed river. The project includes channel widening, a pedestrian path, two bridges, temporary access roads, and relocation of utilities. The area contains a wetland, informal recreational uses, several homes, and a culturally significant riverside location.

Problem

The initial concept reduces modelled flood depth in the central business district but removes part of the wetland, increases construction traffic, and may transfer flood levels downstream. Early public meetings also identify concern about access, noise, and cultural disturbance.

Analysis

The EIS team establishes a catchment boundary rather than limiting analysis to the construction footprint. Hydrology, ecology, traffic, noise, heritage, social impact, and climate specialists use common flood scenarios and receptor maps. The no-project case includes continued development and projected changes in intense rainfall. Three alternatives are assessed: the original channel-widening design, a smaller widening combined with upstream storage, and a distributed “green infrastructure” package with targeted property protection.

The team reports model uncertainty by comparing calibration performance, rainfall scenarios, blockage assumptions, and downstream water levels. The wetland assessment identifies both habitat loss and potential improvement from restoring degraded sections. The heritage assessment recommends seasonal restrictions, exclusion zones, and community-appointed observers during ground disturbance.

Approach and solution

The preferred design combines limited widening with upstream storage and wetland avoidance where feasible. The approval package includes:

• no disturbance within a mapped cultural exclusion zone;
• staged earthworks outside sensitive ecological periods;
• sediment controls with storm-event inspections;
• downstream water-level monitoring during major rainfall;
• a public access plan and grievance mechanism;
• independent ecological and heritage supervision; and
• adaptive triggers requiring design review if monitored downstream levels exceed the agreed threshold.

The EIS clearly states that flood protection improves for the target district but does not eliminate flood risk. Residual wetland loss remains, and the decision-maker requires restoration and long-term maintenance funding.

Outcome

The regulator approves the project with conditions, including additional downstream monitoring and a commitment to revisit the design after three wet seasons. The transparent alternatives analysis improves public confidence even though not all objections are resolved.

Learning points

The case demonstrates that an EIS can change project design, not merely document a predetermined proposal. It also shows the value of catchment-scale boundaries, explicit uncertainty, cumulative analysis, enforceable conditions, and recognition of cultural and distributional effects.

13. Common limitations and best practices

Common weaknesses include inadequate alternatives, baseline surveys conducted in one season, inconsistent assumptions between technical reports, overreliance on model outputs, generic mitigation, unquantified cumulative effects, inaccessible language, and failure to track commitments after approval.

Best practice is to begin with a clear decision question, define boundaries early, use proportionate but defensible evidence, involve affected communities before design is fixed, apply the mitigation hierarchy, test sensitive assumptions, disclose residual effects, and convert commitments into auditable conditions. Independent review is particularly valuable for complex models, high-consequence hazards, and projects with contested evidence.

Chapter Summary

An Environmental Impact Statement is an evidence-based decision document within the broader EIA process. It defines the project and its boundaries, establishes an appropriate baseline, predicts effects through source–pathway–receptor analysis, evaluates alternatives, and proposes mitigation, monitoring, and enforceable conditions. A sound EIS distinguishes measured facts from assumptions and professional judgement, addresses cumulative and induced effects, and communicates uncertainty through confidence statements, scenarios, and sensitivity analysis.

Decision support tools such as MCDA and expected-loss analysis can improve transparency, but they require defensible criteria, weights, probabilities, and safeguards against false precision. Legal standards, precaution, ecological limits, and environmental justice cannot simply be traded away through a numerical score. Effective regulatory review depends on traceability, quality assurance, consultation, accessible communication, and a clear commitments register. Modern tools such as GIS, remote sensing, sensors, and AI can strengthen assessment when their limitations, governance, and data risks are openly managed.

Key Terms

Environmental Impact Statement: The formal document presenting a project’s effects, alternatives, mitigation, and residual impacts for decision-making.

Environmental Impact Assessment: A systematic process for identifying, predicting, evaluating, and managing environmental effects.

Baseline: The existing environmental and social condition used as a reference for assessing change.

Valued Environmental Component: An environmental, social, cultural, or economic feature considered important in the assessment.

Scoping: The process of identifying important issues, boundaries, alternatives, and study requirements.

Impact pathway: The connection from a project activity through an environmental medium to a receptor and consequence.

Receptor: A person, ecosystem, species, resource, or asset that may be affected.

Significance: The importance of an effect considering magnitude, sensitivity, duration, extent, reversibility, and other factors.

Mitigation hierarchy: The sequence of avoidance, minimization, restoration, and offset or compensation.

Residual effect: An effect remaining after mitigation.

Cumulative effect: An effect produced by interaction with other activities or trends.

Uncertainty: Imperfect knowledge about measurements, models, scenarios, systems, or values.

Adaptive management: Structured management that uses monitoring and pre-defined responses to address uncertainty.

Multi-criteria decision analysis: A method for comparing alternatives against weighted criteria.

Approval condition: A legally or administratively enforceable requirement attached to project authorization.

Commitments register: A traceable list of mitigation and monitoring obligations, responsibilities, and verification methods.

Environmental justice: Fair distribution of environmental benefits and burdens and meaningful participation in decisions.

Cumulative effects: Combined effects of the proposal and other past, present, or reasonably foreseeable activities.

Review Questions

5 Short Answer Questions

1. What is the difference between an EIA process and an EIS document?
2. Name four types of boundaries that should be defined in an EIS.
3. Explain the purpose of a commitments register.
4. Identify three important sources of uncertainty in impact assessment.
5. What is meant by residual effect?

5 Long Answer Questions

1. Explain how baseline data, impact pathways, and significance evaluation are connected in an EIS.
2. Discuss the mitigation hierarchy and explain why residual effects must be assessed after mitigation.
3. Describe how multidisciplinary technical studies can be integrated into one coherent EIS.
4. Explain the strengths and limitations of multi-criteria decision analysis for project selection.
5. Discuss the ethical responsibilities of EIS authors and the role of public participation.

3 Analytical / Application Questions

1. A road project is predicted to have low local noise impacts, but three other road projects are planned within the same community. Design a cumulative-effects assessment approach.
2. Construct a monitoring and adaptive-management framework for a construction project whose groundwater drawdown prediction has low confidence.
3. Using the weighted-sum method described in this chapter, compare two hypothetical project alternatives and perform a sensitivity test on at least one criterion weight. Explain whether the ranking is sufficiently robust for a recommendation.

Multiple Choice Questions with Answers

Question 1

Which statement best describes the role of an Environmental Impact Statement?

A. It is a marketing document prepared after construction is complete.

B. It is an evidence-based document supporting environmental decision-making.

C. It is a legally binding permit in every jurisdiction.

D. It replaces all engineering design documents.

Correct Answer: B. It is an evidence-based document supporting environmental decision-making.

Question 2

Which item is an example of an associated project component that should normally be considered in an EIS?

A. An unrelated national policy published decades earlier

B. A temporary access road serving the construction site

C. A private activity with no connection to the project

D. A hypothetical technology with no possible application

Correct Answer: B. A temporary access road serving the construction site

Question 3

What is the correct order in the mitigation hierarchy?

A. Offset, restore, minimize, avoid

B. Minimize, offset, avoid, restore

C. Avoid, minimize, restore, then offset where appropriate

D. Restore, avoid, offset, then minimize

Correct Answer: C. Avoid, minimize, restore, then offset where appropriate

Question 4

In the weighted-sum MCDA method, what condition should the criterion weights normally satisfy?

A. All weights must be greater than 1.

B. Weights must sum to 1 and normally lie between 0 and 1.

C. All criteria must have identical weights.

D. Weights must be expressed in physical units.

Correct Answer: B. Weights must sum to 1 and normally lie between 0 and 1.

Question 5

Using the chapter’s example, what is the weighted score for Site A?

A. 69.75

B. 70.25

C. 73.25

D. 80.00

Correct Answer: C. 73.25

Question 6

Which statement best communicates uncertainty?

A. The model is perfect, so uncertainty is irrelevant.

B. No conclusion is possible whenever data are incomplete.

C. The conclusion, evidence, limitation, and management response are stated together.

D. Uncertainty should be removed from the executive summary.

Correct Answer: C. The conclusion, evidence, limitation, and management response are stated together.

Question 7

What is a residual effect?

A. An effect predicted before any mitigation is designed

B. An effect remaining after mitigation is applied

C. An effect caused only by public consultation

D. An effect that occurs outside the legal boundary

Correct Answer: B. An effect remaining after mitigation is applied

Question 8

Which feature makes an approval condition more effective?

A. It uses broad language with no responsible party.

B. It is measurable, assigned, enforceable, and linked to an impact.

C. It repeats the project vision without performance criteria.

D. It avoids monitoring to reduce administrative costs.

Correct Answer: B. It is measurable, assigned, enforceable, and linked to an impact.

Question 9

Why should the no-project alternative be assessed?

A. It is always environmentally harmless.

B. It provides a reference for future conditions and forgone effects or benefits.

C. It automatically receives the highest MCDA score.

D. It eliminates the need for baseline studies.

Correct Answer: B. It provides a reference for future conditions and forgone effects or benefits.

Question 10

What is an appropriate governance principle when using AI in EIA?

A. AI output should be accepted without professional review.

B. AI should replace public participation.

C. AI outputs require validation, documentation, and human accountability.

D. AI removes the need to disclose data limitations.

Correct Answer: C. AI outputs require validation, documentation, and human accountability.