A year ago, we asked Pinnacle Engineering’s structural engineers how artificial intelligence tools might change their work. Their answers covered what AI could do, the risks it might create, and its potential limits.
This year, we asked a different kind of question. Not what AI could do, but what actually happened when those tools are applied to a structural engineer’s daily work tasks.
We surveyed the team again, asking where AI has saved real time, where it has fallen flat, whether their confidence in these tools has grown or shrunk over the past year, and what specific tools they have actually used.
A year of hands-on use has moved some of those opinions, and not all in the same direction.
How Has AI Actually Saved Time For Structural Engineers?
Last year’s responses were mostly hypothetical. This year, the team had specifics.
Pinnacle Engineering’s President, Tom Moore, uses AI tools for transcribing:
“It saves me time by transcribing recordings so I don’t have to type them. It has fallen flat when it comes to anything involving engineering judgment.”
Structural Engineer Jean Paul Bustamante had a similar experience with writing tasks, but hit a wall on anything code-related:
“It has been helpful for correcting my grammar and spelling when sending emails and writing reports. However, it has fallen short when I needed to find specific engineering codes and standards. In those cases, I ended up relying on my old-fashioned reference books and, of course, seeking guidance from senior engineers in the office.”
Pinnacle Engineering’s Principal and Founder, Kip Ping, has shifted most of his own AI use toward research that used to mean a Google search:
“The place I have been using AI most is for researching that I used to do with Google searches. I find AI is much more adept at quickly finding the answers I am looking for. It hasn’t fallen flat, but I do find the rapid changes that are occurring in the field to be daunting. It seems about the time I think I’m getting a handle on it, everything has changed.”
Structural Engineer Abby Lehmenkuler described a similar shift, calling AI a smarter version of search:
“I use it like a ‘smart Google search.’ Google has become so saturated with promoted websites and irrelevant information that it can be difficult to find what you’re actually looking for. AI can consolidate information from multiple sources into a more coherent and relevant response, while still allowing me to verify the referenced resources for accuracy.”
But every answer had a boundary attached to it. Structural Engineer Nolan Slagle uses AI to get a rough start on county-specific building codes and frost depths, but never treats the answer as final:
“While it’s nice that the AI summaries will sometimes cite their source, it never actually comes from somewhere with authority. It gives me a ballpark answer that I still need to take time to verify.”
He had another example, too, from watching co-ops turn to AI before turning to an engineer:
” One answer it gave that stuck in my head was one that is uniquely technically correct, but utterly useless, answered in such a generic fashion with one solution that only a specific computer program could do, and something that no one would actually do.
The best way I could describe it is if I asked it how to determine where I was in the woods and it answered, ‘just triangulate three satellites in space’, rather than just looking at your map and surroundings to determine location by what you can see.”
Every response from our team drew the same line.
AI tools improve efficiency in administrative tasks, including transcribing, drafting, proofreading, and surface-level research. It doesn’t save time and can create more work for anything that requires structural engineering expertise, including understanding building codes, calculations, and judgment calls that carry real consequences if they’re wrong.
Do Structural Engineers Trust AI More Than a Year Ago?
Where the team largely agreed on where AI’s usefulness ends, confidence in the tools themselves moved in different directions.
Tom’s view hasn’t shifted much:
“I feel about the same. It will never replace us, but it can be a tool to help us.”
Kip has expanded how he uses AI, but treats it the same way he treats any tool he didn’t build himself:
“I have continued to expand the ways I am using it, but also retain healthy skepticism, something that I think should always remain with us. I feel the same way about the software I’ve used for years. Engineers should never fully trust any of the tools we use and should spot-check outputs from any of them. Trust, but verify.”
Abby is the one engineer who reports more confidence than a year ago, though it came from learning the tool’s failure points rather than the tool improving:
“Slightly more confident. From trial and error, I know how to ask AI better questions to get better output. I also know when and where AI will commonly lead me astray to incorrect information.”
Jean Paul remains unconvinced of anything involving calculations:
“For engineering tasks, especially calculations, I don’t feel confident relying on AI. I tried using it once, but I had to spend time teaching it which code and formulas to apply before it could produce the results I needed, so that’s a dealbreaker for me.”
Nolan’s confidence moved the most, and in the opposite direction:
“Truthfully, far less. Ignoring issues with hallucinations, and even AI companies like Microsoft stating that it’s only best suited for cases where deterministic accuracy is not required, the fundamental issue is that any tool that is not clear in how it is doing something and what assumptions it is making is flawed.”
He described what happens once that trust breaks:
“I have dealt with tools that have given me the green check that all is well, and then by happenstance, I see that something is failing. At that point, not only is that terrifying that it’s telling me something works when it doesn’t, but I lose any trust in the program.”
Even the more skeptical members of our team don’t think that AI tools are entirely useless. But how much trust these tools deserve is still an open question, as with any newly introduced engineering software. The consequences of AI tool errors are greater in the structural engineering field, where they can affect a building’s safety.
What Happened When Our Structural Engineers Tested AI Tools at Work?
Beyond general impressions, several engineers described specific tools they have used over the past year.
Tom’s use stayed simple and got simple results:
“I have primarily just focused on using it to make simple tasks go faster, and it has done pretty well.”
Jean Paul kept his use just as narrow, limiting AI to proofreading:
“I used AI tools mostly to check my grammar and my spelling at the time of writing emails or reports.”
Abby piloted a project to test AI’s capabilities in structural engineering. She uploaded past residential inspection reports into ChatGPT for Business and built a tool meant to generate new reports from a few engineering prompts:
“In practice, it did not work well. It largely copied and pasted phrases from other reports in the database rather than producing a coherent technical report with accurate conclusions. What we found is that the AI tool did not replace engineering judgment in that manner. It still required my technical thoughts and conclusions to be developed first before it can effectively help refine the wording, improve readability, and make the report flow better.”
Kip’s next experiment with AI tools is still in progress. He is currently looking into whether AI can support quality assurance and quality control review on construction documents. This potential use case could move AI tools in structural engineering beyond mainly administrative tasks for the team.
AI in Structural Engineering: Frequently Asked Questions
What is AI actually being used for in structural engineering right now?
Based on our team’s experience, AI tools primarily handle administrative work in structural engineering, including transcribing site visit recordings, drafting documentation, proofreading reports and emails, and speeding up early-stage research. It is not yet handling design decisions, code compliance, or structural engineering calculations.
Can AI replace a structural engineer?
Not based on what our engineers have seen. AI tools lack accountability, on-site judgment, and the ability to verify their own output against authoritative codes required for the job. Several engineers described AI giving plausible-sounding answers that turned out to be wrong or unverifiable.
Is AI reliable for structural engineering calculations?
Not yet, according to our team. Using AI tools for calculations currently requires the engineer to already know which code section or formula applies, which limits their usefulness beyond double-checking work the engineer has already done by hand.
About Pinnacle Engineering
Pinnacle Engineering’s wide-ranging expertise and commitment to challenging the norm in structural engineering set us apart. Our clients have been coming back since 1996 for a reason.
Our engineers specialize in commercial, custom residential, and specialized design projects.
If you’re interested in working with us, reach out here.
