Three Lessons For Building Trust In Construction AI
Francesco Iorio is co-founder and CEO of Augmenta, a company automating building design for the construction industry.
gettyArtificial intelligence is taking on increasingly complex work in engineering, manufacturing, transportation and other safety-critical industries. But as the technology advances, verification, certification and regulation are struggling to keep pace.
Generative AI, in particular, needs to be treated as a system operating under uncertainty. Its power lies in its ability to rapidly explore countless ways to answer a question or solve a problem, considering far more possibilities than a human could on their own. But greater possibility also introduces uncertainty. To fully maximize the value of generative AI, we need to first focus on validating and verifying the processes that produce results, and then the outputs themselves.
We’ve seen this tension between technological capability and trust play out before, and autonomous driving offers a useful roadmap for what construction and other industries where technology has real-world, physical consequences can expect.
Every transformative technology has faced a “trust gap” before widespread adoption. Most people assume that as technology improves, trust will follow and oversight gradually lessens. Autonomous driving has shown that the reality is more complicated.
As autonomous vehicles became more capable and moved from controlled testing environments onto public roads, they attracted greater scrutiny. The closer the technology came to operating independently, the more questions arose around validation, liability and regulation. In other words, new technology did not eliminate the need for oversight; instead, it raised the stakes for proving that the technology should be trusted.
Construction has a rare opportunity to learn from that trajectory rather than repeat the same mistakes. As AI takes on more complex work, we need to simultaneously create the systems and standards to validate its output and define the role of human accountability.
That starts with rethinking how we build the technology, the role we ask people to play and the standards the industry sets for itself.
Verification, or ensuring that an AI system’s processes and outputs perform reliably within defined constraints, shouldn’t be an afterthought or left solely to regulators. If we want AI to take on more complex work, we need to move beyond verifying individual outputs to validating the processes that create them. That means understanding where uncertainty or error can occur, how it can be controlled and whether these systems are reliable and repeatable enough to trust as they become part of how we design, build and make decisions every day.
Other industries offer a precedent. Aerospace has long relied on rigorous certification processes for the software and manufacturing methods used to build aircraft, while the pharmaceutical industry has developed frameworks for validating computational models used in drug development. Construction will need its own AI validation infrastructure.
In an industry as safety-critical as construction, leaders cannot afford to assume AI will produce the right answer. Trust will come from building processes that consistently identify and catch errors before they have serious consequences.
The goal for AI in construction shouldn’t be removing engineers, designers, contractors or other skilled professionals from the process. It’s allowing them to spend less time on repetitive tasks, work more efficiently and expand their capacity at a time when project demands are growing and skilled labor remains in short supply.
AI can dramatically reduce the time required to generate or explore an idea, but it cannot yet shorten the time required to validate it. That makes human expertise and institutional knowledge even more valuable because they define what quality looks like and provide the judgment needed to recognize when something is wrong.
As automation expands, engineering, design, operations and other roles will evolve in parallel. For example, rather than producing every element themselves, engineers can spend more time setting goals and rules, evaluating alternatives and validating results. For the foreseeable future, accountability will still belong to the people willing to put their names behind the work.
Waiting for regulation means reacting instead of leading. It’s no secret that AI capabilities are evolving quickly, and regulators face the difficult task of developing standards for ever-changing technology across countless industries.
That’s why industry leaders should work together to establish best practices around validation, documentation and sign-off now. We don’t need to solve every question about AI standards immediately, but we do need a foundation for determining where and how these systems can be rigorously tested and validated.
If our industry doesn’t define what trustworthy AI looks like, someone else will. By taking the lead, we can help build trust across construction while making responsible AI deployment a competitive advantage for the companies that get it right.
In construction, trust has always been earned through reliability, consistency and accountability, and that shouldn’t change with AI. Construction leaders will choose technology partners that make their work more efficient without compromising the rigor that protects their projects and businesses.
As AI capabilities advance across the industry, the companies that succeed will be those that can demonstrate reliability, consistency and accountability. The ultimate test is simple: Are people willing to put their names behind the work their technology produces?
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
