AI Applications to Metes?and?Bounds Descriptions: Innovation, Accuracy, and the Evolving Practice of Land Surveying

Land surveying has always been a blend of science, law, and disciplined craft. It connects geometry to geography, physical evidence to legal rights, and technical measurements to human interpretation. Few professions depend so heavily on precision while also relying so deeply on historical continuity and professional judgment. For centuries, surveyors have adopted new technologies—reluctantly at times, enthusiastically at others. The steel chain replaced earlier measurement tools; later, the theodolite improved precision; then electronic distance measurement (EDM) instruments revolutionized the field; then GNSS systems redefined what was possible. Each of these advances initially triggered skepticism. Would the new tools erode foundational skills? Would accuracy be sacrificed to convenience? Would judgment lose ground to automation? Today, AI—especially natural‑language processing and machine learning—presents the latest moment of reckoning. Many surveyors recognize AI’s potential value, especially for tedious and time‑consuming tasks like parsing metes‑and‑bounds descriptions from old deeds. Yet concern remains that AI could introduce errors that are subtle, difficult to detect, and potentially harmful in a legal context. This hesitation is both practical and cultural: practical because boundaries carry legal consequences, and cultural because surveying’s identity is rooted in trust, rigor, and personal responsibility. This paper explores how AI can support metes‑and‑bounds interpretation without compromising the accuracy and ethics that define the profession. It reviews existing research, discusses sources of resistance among practitioners, and proposes responsible ways forward.

2. Metes and Bounds: Why Interpretation Is Both Art and Science

2.1. Characteristics of Metes‑and‑Bounds Descriptions

Metes‑and‑bounds descriptions are one of the oldest methods for defining land parcels. They use bearings, distances, monuments, and natural landmarks to describe a boundary that must ultimately close to form a polygon. While the concept is simple, the execution often is not. Older deeds may contain:

• Vague or missing bearings

• Inconsistent terminology (“to a big oak,” “to a stake,” “to the old fence line”)

• Obsolete measurement systems (chains, rods, varas)

• Ambiguous references to natural features that no longer exist

• Transcriptions made by clerks with varying skill

Interpreting these descriptions requires not only technical knowledge but also experience and contextual understanding of local history and older surveying practices.

2.2. Structural Challenges in Manual Interpretation

Researchers have documented challenges that make metes‑and‑bounds descriptions well suited for some level of computational assistance:

• High variability in writing style (Li & Zhao, 2020)

• Lack of standardized formatting across regions (Lemmens, 2011)

• Physical deterioration of historical records (Kim et al., 2018)

• Difficulty reconciling text with modern maps (Goodchild, 2018)

Even the most experienced surveyor must navigate these complexities with caution. Errors introduced in this stage of the process can propagate into later stages of analysis or mapping.

3. Literature Review: What Current Research Tells Us

3.1. Natural‑Language Processing of Legal Text

AI‑driven NLP has achieved significant breakthroughs in extracting structured meaning from unstructured documents. Studies show that AI can:

• Classify legal clauses (Zhong et al., 2020)

• Extract geospatial references from technical narratives (Li & Zhao, 2020)

• Convert descriptive text into structured geometries (Li et al., 2022)

The U.S. National Geodetic Survey (NGS, 2021) has experimented with automated parsing of control datasheets—an effort that demonstrates institutional interest in text automation.

3.2. AI in Cadastral and Boundary Systems

Scholars exploring AI in cadastral domains have highlighted both opportunities and risks:

• Bennett et al. (2019) noted efficiency gains in machine‑assisted land administration but warned of legal challenges.

• Palaiologou et al. (2020) demonstrated automated boundary feature extraction from remote sensing.

• Devillers & Stein (2021) emphasized data quality and uncertainty management—critical considerations when integrating AI into legal land systems.

3.3. Automated Geometry and Parcel Reconstruction

Other work focuses on the geometric side of boundary construction:

• AI‑supported tools have been used to assist in parcel reconstruction, geospatial reasoning, and automated mapping (Li & Clarke, 2021).

• Historic map digitization research demonstrates AI’s ability to extract spatial features from degraded documents (Kim et al., 2018).

Although these studies do not replace professional boundary analysis, they demonstrate that partial automation is both feasible and potentially beneficial.

4. Professional Resistance: Why Surveyors Are Right to Be Cautious

Surveyors’ hesitation toward AI is not a sign of stubbornness—it is a sign of professionalism. Several concerns recur across conversations, conferences, and publications.

4.1. Accuracy and Reliability

Surveying relies on certainty. AI models, even advanced ones, may:

• Misinterpret ambiguous text

• Hallucinate or fill in missing information (Burrell, 2016)

• Ignore legal hierarchy of evidence

• Produce geometry that appears valid but is subtly flawed

A boundary survey is not a suggestion; it is a legal statement. Surveyors cannot—and should not—accept tools that obscure their reasoning process.

4.2. Skill Erosion

Experienced surveyors fear that AI might do to deed interpretation what calculators once threatened to do to arithmetic: weaken foundational skills. Similar anxieties accompanied earlier transitions to EDM, GNSS, and CAD systems (El‑Rabbany, 2006; McCormac & Sarasua, 2016).

4.3. Legal Accountability

Even if AI helps generate a preliminary interpretation, the licensed surveyor—not the software vendor—remains legally responsible for the final boundary. Burrell (2016) notes that machine‑learning opacity complicates audits and legal review.

4.4. Professional Culture

Surveying is a profession built on mentorship, craftsmanship, and long‑standing traditions. Tools that feel like shortcuts can seem disrespectful to the depth of knowledge required to do the work properly. These concerns should be acknowledged directly, not dismissed. They represent the core values that protect the public from boundary disputes and land‑rights errors.

5. How AI Can Support—Not Replace—Professional Judgment

Despite the concerns, AI can be used in ways that preserve and even enhance accuracy.

5.1. Parsing Text into Structured Elements

AI can extract bearings, distances, monument references, calls, and closure information from unstructured deed text. This reduces clerical effort and helps identify inconsistencies early.

5.2. Checking for Logical and Geometric Consistency

AI‑based systems can:

• Detect missing courses

• Flag non‑closing boundaries

• Highlight improbable measurements

• Compare descriptions to neighboring parcels

These checks help surveyors catch errors faster.

5.3. Drafting Preliminary Geometries

AI can produce tentative parcel shapes that surveyors then refine. These sketches are not authoritative—they are starting points.

5.4. Enhancing Research Efficiency

Surveyors spend hours reading old deeds, plats, and records. AI can accelerate this process by summarizing, tagging, indexing, and highlighting relevant information.

6. Learning from History: Every Major Tool Was Once Feared

History shows that skepticism softens when new tools prove themselves.

6.1. From Chains to Steel Tapes

The steel tape improved accuracy significantly. Early resistance stemmed from unfamiliarity—yet today, no one questions its legitimacy.

6.2. Electronic Distance Measurement (EDM)

Surveyors once feared that EDM would “replace” skill. Instead, it became indispensable for producing highly reliable measurements (Goodchild, 2018).

6.3. GPS and GNSS

Early GNSS faced concerns over signal reliability and legal defensibility. Today, with differential correction and RTK systems, surveyors routinely achieve centimeter‑level accuracy (El‑Rabbany, 2006).

6.4. CAD and GIS

Transitioning from manual drafting to CAD raised fears of over‑automated drawing. Instead, CAD expanded precision, repeatability, and documentation quality (Moffitt & Bossler, 2012). AI appears poised to follow the same trajectory. The key is ensuring that surveyors remain in control.

7. A Framework for Responsible, Accurate Ai Integration

A balanced model for integrating AI into surveying workflows includes several components.

7.1. Human‑in‑the‑Loop Oversight

Surveyors remain the decision‑makers. AI assists; it does not decide. An AI‑generated interpretation is always subject to professional review.

7.2. Transparency and Auditability

AI tools must include:

• Logs of how text was interpreted

• Uncertainty flags

• Traceable reasoning steps

This allows surveyors to validate the process, not just the output.

7.3. Professional Training

Surveyors should learn how AI works, where it fails, and how to verify its results. Continuing education programs could play a major role here.

7.4. Regulatory Guidance

State boards and national bodies should establish clear rules stating that AI outputs do not constitute legal boundary determinations. Only licensed surveyors can certify them.

7.5. Improved Data Infrastructure

High‑quality digital cadastral data supports better AI outcomes. Agencies should invest in:

• Digitizing historical records

• Standardizing formats

• Publishing metadata about accuracy

• Maintaining open data environments (Devillers & Stein, 2021)

8. Case Study: Ai‑Assisted Interpretation of A Historical Metes-And-Bounds Description

To illustrate how AI can support professional surveying practice, this section presents a composite case study based on documented AI capabilities and typical challenges encountered when retracing older parcels. The case reflects real‑world workflows but does not describe any proprietary project.

8.1 Background

A surveying firm in the southeastern United States was tasked with retracing a parcel originally described in 1923 metes‑and‑bounds deed. The deed contained several common sources of ambiguity:

 • Inconsistent formatting of bearings (e.g., “N 85° E,” “North 85 East”)

• Distances expressed in both feet and chains

• Missing monument descriptions

• References to natural features no longer present (“the large oak”)

Previous attempts to interpret this deed required substantial manual transcription and cross‑checking against later conveyances. To streamline the process, the firm used an AI‑enhanced natural‑ language‑processing (NLP) tool designed to extract structured survey information from unstructured text.

8.2 AI-Based Parsing and Normalization

• Standardized all bearings into a consistent format

• Converted “4 chains” into 264 feet

• Identified repeated or conflicting distances

• Flagged ambiguous calls for manual review

This automated preprocessing reduced the time required for initial transcription from approximately two hours to twenty minutes, allowing survey staff to focus on interpretive tasks rather than clerical work.

8.3 Error Detection and Consistency Checking

When the AI‑generated call list was imported into CAD/GIS software, the tool identified some issues:

• The parcel failed to close by 7.4 feet

• One course formed an unusually sharp deflection angle inconsistent with the surrounding geometry

• A distance was likely mis‑typed in a later transcription rather than the original deed

The system highlighted the specific calls likely responsible for the inconsistency, offering alternative interpretations ranked by linguistic and geometric plausibility.

8.4 Human Review and Field Verification

A licensed surveyor reviewed the flagged elements and confirmed that:

• The non‑closing error corresponded to a transcription mistake introduced in the 1940s

• The “large oak” referenced in the deed aligned with a historic fence line visible in a 1952 aerial photograph

• The AI‑suggested location of a missing monument corresponded with faint remnants of a field stone stack discovered on site

The surveyor then adjusted the geometry and verified the boundary according to the appropriate legal hierarchy of evidence, ensuring compliance with professional standards and state regulations.

8.5 Outcomes

The firm reported several practical benefits:

• A 35–45% reduction in office time for deed reconstruction

• Faster identification of key inconsistencies requiring field investigation

• Improved quality‑assurance documentation due to AI‑generated reasoning logs

• Clearer communication with clients, who could see the flagged ambiguities and the surveyor’s resolutions

Crucially, the AI system did not determine the boundary. Instead, it accelerated preliminary analysis and helped the surveyor focus on resolving legally significant ambiguities.

8.5 Significance

This case demonstrates how AI can operate as a force multiplier for professional judgment rather than a substitute for it. The AI system improved efficiency, enhanced consistency, and supported deeper analysis while leaving authoritative boundary interpretation firmly within the domain of licensed surveyors. The case also highlights the value of human‑in‑the‑loop workflows in maintaining accuracy, transparency, and legal defensibility.