Dr. Neda Naghshbandi’s Post

I was delighted to attend the Institution of Civil Engineers (ICE), and Bentley Transportation Solutions, AI Roundtable on 26 March, engaging with industry leaders who are committed to defining a pragmatic, delivery‑focused AI strategy for infrastructure. One of the topics we discussed was what has changed over the last six months. In my view, the biggest shift has been our definition of #success. Success is no longer about building a good model; it’s about deploying AI that works in a #live #infrastructure environment and can #scale. We’re recognising that scaling AI is less a technical challenge and more a #governance and #integration challenge. As a result, we’re placing equal emphasis on #leadership ownership, #assurance frameworks, #safety, and #repeatability, not just the technology itself. Accordingly, the focus has moved away from isolated pilots or proving that AI can work. Instead, we’re tackling real operational problems and building the full AI operating model end to end; across #dataflows, #workflows, #governance, #assurance, and structured #human oversight. It’s a delivery‑focused approach designed to generate real operational value. One of the big #themes that keeps coming up is a leadership tension between AI #believers and #nonbelievers. That balance is starting to shift, with more leaders leaning in, experimenting, and learning what really gets in the way. Anyway, experimentation on its own isn’t enough. AI needs to be #owned as a core part of how we deliver, not treated as a side project, which means clear leadership accountability. And the same challenges keep surfacing: 1-fragmented delivery across organisations, 2-gaps in trust and skills, and 3-procurement models that simply aren’t built for iterative, evolving AI. Julie Wood Mark Coates FCIHT, FCInstCES Anne-Marie Friel Yeunjin (YJ) Kim Amanda Leness Darren Silvester James Daniel FCIAT FCIHT James Crumly Lydia Walpole Guy Beaumont