Las Vegas — Trimble Dimensions 2024, held in the buzzing halls of the Venetian Convention Center, was a showcase of transformative technology for the built environment. Amid the digital displays and technical presentations, I had the pleasure of meeting with Karoliina Torttila, Trimble’s Director of AI, for an in-depth conversation. Torttila is Trimble’s quiet powerhouse in harnessing artificial intelligence to meet the nuanced needs of architecture, engineering, and construction (AEC). Our discussion underscored the practicality driving Trimble’s AI initiatives, turning AI from an abstract idea into a tool that supports real-world challenges.

Karoliina Torttila

Torttila greeted me warmly, her passion evident as she described her work. “Our job,” she said, “is to adapt AI to make it genuinely useful for our industry.” In Torttila’s view, AI must address the unique challenges of AEC, where construction sites, structural materials, and surveying equipment require specialized knowledge that traditional AI models simply don’t have. “There are fantastic general AI models, but they aren’t designed to handle the complex, ever-changing environments of construction and surveying.”

One of the most significant hurdles Torttila faces is training AI to recognize the highly specific visuals and conditions on construction sites. While AI is adept at recognizing common objects in everyday settings, it falters when applied to niche environments like electrical substations or construction zones, where the visual data is dramatically different. Torttila explained how her team at Trimble refines these models to interpret everything from survey instruments to steel beams in real-world conditions, filling in gaps left by standard AI training sets. “We’re not building a ‘ChatGPT for construction,’” she laughed. “Instead, we’re refining existing models to make them reliable in our specialized settings.”