From Unscripted with Jeff Pedowitz, featuring Nico Ianeselli, civil engineer and Senior Director of Technology at Chen Moore and Associates

Most of the AI conversation lives in software. Nico Ianeselli works in concrete and steel. He's a civil engineer and technology executive with 13+ years bringing AI and digital twins to airports, bridges, dams, and water systems, including some of the largest infrastructure projects on the planet, and he's now building an AI practice inside a civil engineering firm.

Jeff wanted to know what AI actually does once it leaves the screen and has to hold up a bridge. Here's what came back.

1. "Human in the lead," not human in the loop

Ianeselli used to think engineering was just slow to adopt technology. Then he reframed it: in his field, an overlooked error isn't a bug you patch. It's a structural foundation that fails, and people die. That's why his standard for infrastructure AI goes one step past the industry cliche.

"There is the absolute need to have human in the lead, not in the loop, for this kind of industry."

The distinction matters. In the loop means a person reviews AI output. In the lead means the person owns the decision and AI serves it. For any business where being wrong is catastrophic (engineering, medicine, finance), that's the governance model worth stealing. He pairs it with a second idea: forensic preservation for AI, the ability to freeze a system in stasis and investigate exactly why it made a decision before it ever makes it again.

2. Europe is crawling, America is walking, China is running

Fresh off a conference trip to China, his geopolitical read is blunt. Europe over-regulates first and U-turns later, losing years. America debates, keeps the human in the center, and moves at a walk. China has one way: beat the US in AI, whatever it takes, and even engineers in the trenches see little of what's happening outside.

His cold engineering assessment: he admires the pace (the robots and autonomous vehicles are astonishing) while preferring the human-centered debate at home. The unlock for the West, in his view, isn't matching China's speed. It's education that builds enough trust to stop relitigating settled questions, like whether autonomous systems can be demonstrably safer than humans with a 1.5 second reaction time.

3. Digital twins are already paying for themselves

Forget the SimCity comparison. A digital twin is a live virtual replica of a physical asset, synced to reality, where you can run what nothing physical survives: hurricanes, fires, earthquakes, a bridge fully loaded with stopped traffic after an accident. Add the fourth dimension, time, and you can watch a building age 50 years or flip from residential to commercial and see what happens to occupancy and energy.

This isn't a pitch deck. On the Muskrat Falls hydroelectric project, one of the biggest dams in North America at $2.2 billion, his team used 4D simulation to sequence concrete pours, coordinate cranes across multiple work areas, and keep workers out of risk zones. His words: it avoided major mistakes. Meanwhile the bridge near his home in Miami is 9 years behind schedule, a reminder of what unsimulated megaprojects cost.

The forward case is even bigger. Engineers design for worst-case scenarios plus safety factors that are always overestimates. AI, and eventually quantum computing, can compute the real worst case from all variables in real time. Shave a bridge deck from 20 feet of material to 15 and the capex savings are enormous.

4. The answer to crumbling infrastructure isn't rebuilding. It's sensors.

America can't tear everything down and rebuild like Dubai. Ianeselli's alternative: embed IoT sensors in existing bridges and buildings, connect them to digital twins, and replace random cyclical maintenance with intervention exactly when tolerance levels drop. Green, yellow, orange, red. Send the crew at yellow, never see red.

The payoff he cites: extending a bridge's standard 50-year lifespan toward 200 years. Fixing early is cheaper than stitching the sock when it's full of holes. That's predictive maintenance as public policy, and the same logic applies to any expensive asset your business runs, physical or not.

5. The energy debate is measuring the wrong number

Asked how he reconciles building data centers with the human side of AI, Ianeselli flips the framing. On water: mass is conserved, cooling water becomes vapor and comes back as rain, it doesn't vanish. On energy, he borrows from Jensen Huang's figure that AI uses roughly 4% of grid energy and adds his own twist:

"I'm not worried about the 4%, because that's maximum efficiency. I'm worried about the 96% that's not using it."

Drive through downtown Miami at 2 a.m. and every office tower is lit with nobody inside. AI-run buildings that switch themselves off, HVAC that learns occupancy patterns, AI designing its own data centers with new cooling fluids and new steel alloys: the waste AI can eliminate dwarfs the energy it consumes. Whether or not you buy the full argument, it's the strongest version of the case, and it's the one your skeptics haven't heard.

The liability question nobody can answer

Would he let AI sign off on a design with no human involvement? Not today, and the reason is a 100-year-old institution: the Professional Engineer, created around 1920, whose stamp makes a named human liable for every drawing. If an AI-designed building collapses, who's liable: the programmer, the CEO, the model vendor? "There's no answer," he admits, and until there is, the human stays in the lead. He does think the line will blur. It was ink-sharp in the past. It's already going gray.

His 50-year vision runs further than most guests are willing to go on the record: AI that stops emulating human data and starts contributing new engineering principles, the way the arch once changed everything. He'll take the floating mountains from Avatar as a design brief. "If we don't push the limit, we will never know which limits we have."

FAQ

Who is Nico Ianeselli? Nicola Ianeselli is a civil engineer and Senior Director of Technology at Chen Moore and Associates, with 13+ years integrating AI, digital twins, and blockchain into civil, aviation, and marine projects, including Mexico City's airport and the Muskrat Falls Hydroelectric Project. He holds dual master's degrees in engineering and architecture.

What is a digital twin? A virtual replica of a physical asset, synced to the real thing, used to simulate events (storms, fires, load spikes, aging over decades) at minimal cost. With physical properties embedded, it's a working engineering model, not just a visualization.

Where are digital twins actually used today? Bridge vibration monitoring, construction sequencing on megaprojects like Muskrat Falls, and predictive maintenance on infrastructure. Beyond that, they're still more pitch than practice.

What is "human in the lead"? Ianeselli's stricter version of human-in-the-loop for safety-critical industries: the human owns the decision and remains liable; AI informs, simulates, and accelerates, but doesn't sign.

What's the main business takeaway? Match your AI governance to your cost of error, chase the waste AI can eliminate before debating the energy it consumes, and instrument your existing assets before you replace them.

Unscripted with Jeff Pedowitz features honest conversations with remarkable people from security and science, business and medicine, technology and the arts. Subscribe wherever you get your podcasts.

The Pedowitz Group | pedowitzgroup.com