As AI raises the appearance of capability, AEC leaders must get better at identifying real competence.
Over the past year, I have watched the appearance of competence change inside our industry. I see it most clearly in my inbox. Entry-level resumes are arriving cleaner, sharper, and more polished than they used to. They are also arriving more interchangeable. AI is helping candidates write to the same template, hit the same keywords, and produce the same well-formatted summaries of the same kinds of experience. The output looks better, but the candidates have not become more distinct. The same pattern shows up in proposals, project summaries, and meeting follow-ups.
None of this is bad on its own. AI helping a graduate produce a stronger resume is fair use of the tool. The leadership problem is what comes next: how do we tell the difference between work that is AI-polished and competence that has actually been earned?
How AI is changing the appearance of competence
This is a measurement problem before it is a talent problem. AI can raise the appearance of capability faster than people can earn the capability underneath. The first draft, the resume, the proposal, the interview answer, all of it now arrives more polished than it would have two years ago. Underneath the polish, the gap between access to knowledge and earned competence is widening, not closing.
The class of 2031 is currently in high school, having entered after ChatGPT launched in late 2022. By the time they reach the AEC workforce, AI-assisted work will feel like the water they learned to swim in. They will not arrive less prepared than prior cohorts. They will arrive looking more prepared than they have any right to look.
That is harder to navigate than it sounds. AEC is not built on clean prompts and clean data. It is built on incomplete drawings, owner pressure, field constraints, and conflicting specifications. AI will polish the deliverables. It will not stand on the slab on a Friday afternoon when the concrete pour is going wrong.
In AEC, that kind of knowing is not abstract intelligence. It is material intelligence: the earned understanding of how the built world behaves under stress, time, chemistry, workmanship, weather, and consequence. The engineers who can articulate it are the ones AI will struggle to commoditize.
What earned competence looks like in an AI-driven workplace
Four habits separate earned competence from AI-amplified appearance:
- Question architecture. The signature skill of the next decade is not giving better answers. It is asking better questions. The engineer who can frame the problem (design issue, constructability issue, risk allocation issue, or field judgment call) before asking AI to solve it shows the first evidence of real competence.
- Stress-testing AI outputs. The professionals who matter most in 2031 will recognize when a confident output is wrong, when a model hallucinates a code reference, and when a generated calculation skips a load case no checker would have missed. Skepticism is a teachable practice, not a personality trait.
- Stakeholder fluency. AI cannot read the silence in a project meeting. It cannot tell when an owner is bluffing about a budget or when a junior engineer is too intimidated to flag a real concern. Reading those cues and translating them into decisions is work AI cannot do well in 2031.
- Adaptive confidence. Technical fluency will need to be renewed constantly, not acquired once. The professionals who thrive will hold conviction loosely, learn fast, and recover from being wrong without losing their footing. That blend is rare, and it is teachable, but only by leaders who already have it themselves.
The equation underneath all four is simple. AI brings speed. Material intelligence brings the evidence. Human intelligence decides what to trust, what to question, and what to do next. (AI + MI) × HI. Firms have to build the multiplier before the polish makes everyone look like they already have it.
How AEC firms can develop real professional judgment
So what does a firm leader do with this on Monday morning? Three things, none requiring a new technology budget:
- Redesign early-career development around exposure, not production: put new hires in front of clients earlier, bring them into forensic investigations, and let them sit in on the difficult conversations.
- Interview for reasoning under uncertainty, not polish: every interview should include a live problem the candidate has not seen before.
- Train managers to develop people, not just review output: teach young professionals to frame the problem, challenge the answer, and explain the risk.
Every generation has worried that the next one would not be ready. The generation that taught us worried about us. We turned out fine. The next generation will too. That means changing how we hire, train, mentor, and verify professional judgment. Our job is not to protect the industry from the class of 2031. Our job is to build firms worthy of them. The smarter you is not a threat. It is an invitation. The only question is whether you are ready to meet the smarter you.
 |
Jigar Desai, PhD, PE, MBA, is a principal and regional manager at ECS Limited, where he leads building science and infrastructure teams across six offices in the Mid-Atlantic. Connect with him on LinkedIn. |
