Would You Cross That Bridge? AI and the Trust Problem in Infrastructure
Every time you drive over a bridge, turn on a tap for a glass of water, or flip a light switch, you are placing your trust in a complex system of infrastructure. You trust that the engineering was sound, the materials were correct, and the analysis was precise. This trust is the silent, foundational contract upon which modern society is built. It is also why the conversation about AI in infrastructure must be fundamentally different from any other domain. The rise of generative-AI has been astonishing, but it has also introduced the concept of “AI slop”āoutputs that are plausible-sounding but often imprecise, inconsistent, or flat-out wrong. In many fields, this is an acceptable tradeoff for speed and creativity: a door the wrong shade of blue may offend a design eye, but it wonāt physically hurt anyone. The reality of the infrastructure sector is that there is no room for approximation, let alone hallucination. A model that is 90% right is a useful start; a structural analysis that is less that 100% right is a catastrophic liability.āÆWould you drive across a bridge that is āhopefullyā designed right? The gold standard of AI in civil engineering is not to provide a creative