The short version

In a trustee’s column for *New Civil Engineer*, AI doesn’t arrive as a wrecking ball for the profession. It shows up more like a new graduate on the project team: promising, fast, occasionally overconfident—and absolutely in need of supervision.

The trustee’s view of AI in engineering is measured. AI tools are treated as extensions of existing construction technology, not a replacement for engineering judgment. From design support to checking calculations, the message is simple: automation can help shoulder the routine workload, but responsibility for safety and performance still sits squarely with human engineers.

AI can assist with engineering work, but it cannot carry professional accountability for the outcomes of that work.

Why this matters on real projects

On paper, AI in construction sounds like a clean upgrade: faster options studies, automated clash checks, smarter scheduling. The trustee’s perspective reminds us that the reality is messier—and more human.

The column frames AI as a continuation of tools engineers already trust: spreadsheets, finite element packages, BIM platforms. What’s new is the scale and speed. An AI model can scan design data, past projects and codes to propose options or flag anomalies in seconds. But the trustee is clear: **suggestions are not decisions.**

On a live project, that distinction is everything.

Imagine a bridge design where an AI assistant highlights members it thinks are over‑designed and proposes slimmer alternatives. That’s useful automation. But the person signing the drawings still has to understand load paths, local conditions and construction sequence—and be ready to defend the design in a safety review or court.

The trustee’s view pushes back against two common extremes:

• The fantasy that AI tools will soon design and sign off whole structures.
• The fear that any use of AI is reckless or unethical.

Instead, the column sketches a middle lane. AI in engineering can:

• Help compare design options faster.
• Support checking and documentation.
• Capture and surface lessons from past projects.

But it **cannot**:

• Take legal or moral responsibility for public safety.
• Replace the need for competent, chartered engineers.
• Understand context the way a site‑hardened engineer does.

For contractors and consultants, that has practical implications. Rolling out AI in construction isn’t just an IT project; it’s a governance project. Who reviews AI‑generated content? How is it recorded in the design file? When does an AI suggestion become part of the formal design basis? The trustee’s emphasis on accountability hints that these questions will matter as much as the algorithms.

What to watch next

• **Codified guardrails for AI use**: Expect more firms and institutions to issue guidance on when AI tools may be used in design, checking and documentation—and how their outputs must be reviewed.
• **Training that pairs judgment with automation**: Professional development will likely shift toward teaching engineers how to interrogate AI outputs, not just how to generate them.
• **Audit trails inside construction technology**: Design and project platforms may start logging when AI was used, what it produced and who approved it, to preserve accountability.
• **Scope creep of "routine" tasks**: As automation improves, the industry will need to keep renegotiating which tasks are safely delegated to AI and which must remain firmly in human hands.
• **Ethics and liability debates**: Professional bodies are poised to wrestle with where responsibility sits when AI‑assisted work contributes to a failure—even if the trustee’s stance is that accountability cannot be offloaded.

Field note from the editor

Reading this trustee’s take, I’m struck by how familiar it feels. Every major shift in construction technology—hand calculations to software, drawing boards to CAD, 2D to BIM—came with the same tension: relief at new efficiency, anxiety about what might be lost.

What’s different with AI tools is the illusion of understanding. A beautifully worded design note or a neat load path sketch from an AI system can look more authoritative than it is. The trustee’s insistence on human accountability is a useful anchor.

If you work in delivery, design or asset management, the real question isn’t whether AI will arrive on your projects—it already has. The question is how deliberately you’ll frame it: as a black box to be feared, or as a fast but fallible junior whose work you always double‑check.

For now, the trustee’s view offers a simple rule of thumb: let automation speed the work, but never let it own the risk.