AI in Civil Engineering: What's Real vs What's Noise

Key Concepts

• AI in Civil Engineering: The integration of artificial intelligence to streamline workflows, verify standards, and summarize complex regulatory data.
• Engineering Judgment: The essential human oversight required to validate AI outputs, manage risk, and ensure quality.
• Desktop Review: An early-stage project phase using AI to analyze site constraints (zoning, environmental, flood maps) before committing to expensive field work.
• Value Engineering: Using AI to explore cost-effective design alternatives and material substitutions.
• Unstructured Data: Large volumes of disparate information (reports, emails, historical project files) that AI can synthesize into structured, actionable summaries.
• Hallucinations: The tendency of AI to occasionally generate inaccurate or fabricated information, necessitating human verification.

---

1. AI Adoption and Workflow Transformation

Ron Lazat, Senior Principal at Kier’s Engineering and Design, emphasizes that AI is the latest in a series of industry disruptors, following the transformative impact of Autodesk’s Civil 3D and BIM.

• Workflow Efficiency: AI is currently being used to automate the verification of documents against industry standards and to summarize complex regulatory requirements.
• Multidisciplinary Coordination: In land development, where civil engineers act as "quarterbacks" for architects, geotech, and traffic engineers, AI helps cross-check plans against various technical reports and specifications to ensure consistency.
• Speed: AI significantly reduces the time spent on research and data retrieval, turning 20–30 minute manual searches into sub-minute tasks.

2. Strategic Application in Project Phases

• Site Selection & Due Diligence: Before investing in costly geotechnical or survey work, firms use AI to perform "desktop reviews." By querying zoning, tax maps, and environmental constraints, AI helps determine if a site is viable, potentially saving clients from pursuing "dead deals."
• Value Engineering: AI assists engineers in comparing material costs and construction methods (e.g., different storm water systems or pipe materials) to optimize project budgets.
• Historical Data Mining: Large firms with multiple offices can use AI to query internal archives of past projects. This allows teams to learn from previous site-specific challenges, such as environmental issues or jurisdictional hurdles, providing a competitive advantage during the bidding process.

3. The Role of Human Judgment

A central theme of the discussion is that "the most dangerous engineer is one equipped with AI but lacking experience."

• Validation: Because AI can produce different outcomes based on the platform or parameters used, experienced engineers must act as curators. Lazat cites a case study where two AI platforms provided cost estimates for a roadway project with a $250,000 discrepancy. The higher estimate was more accurate because it accounted for real-world variables like site clearing, erosion control, and contractor contingencies—factors the AI might overlook without expert guidance.
• Contextual Awareness: AI lacks the "taste" and "judgment" of a seasoned professional who can look at an aerial site map and immediately identify potential flood hazards or access issues.

4. Scaling AI Across Large Organizations

Scaling AI in a firm with 90+ offices requires a structured approach:

• Start Simple: Avoid getting overwhelmed by the sheer number of AI tools (ChatGPT, Copilot, Gemini). Prioritize specific, role-based workflows.
• IT Safeguards: A critical challenge is protecting proprietary information. Firms must establish clear "guardrails" to ensure that internal data is not exposed or misused when querying AI models.
• Controlled Experimentation: Encourage teams to test AI in safe, controlled environments to understand its capabilities and limitations before full-scale implementation.

5. Notable Quotes

• "How do we build a better mousetrap? No matter what you’ve done five years ago or even last year, there’s always ways to improve." — Ron Lazat
• "The most dangerous engineer equipped with AI is an experienced one." — Host (Nick)
• "We’re the trusted advisor... that’s where the expertise comes in." — Ron Lazat

Synthesis and Conclusion

The integration of AI into civil engineering is not about replacing the engineer, but about augmenting their ability to manage data and streamline repetitive tasks. The most successful firms are those that use AI to synthesize large amounts of unstructured data—such as regulatory codes and historical project reports—while maintaining rigorous human oversight. By focusing on "building a better mousetrap" through controlled, strategic adoption, engineering leaders can improve project outcomes, save clients time and money, and maintain the high quality of work that defines the profession.