As mentioned in Part Two: AI in Shipbuilding – Move Fast AND Smart? AI is coming to shipbuilding faster than any technology we’ve seen before. While this rapid pace brings real opportunity, it also introduces challenges that haven’t received much attention. I want to focus on two in particular.
The first is data. Much of the data we have today is either difficult for AI models to access or structured around how we used to work, not how we want to work going forward. If we train AI on information rooted in legacy processes, we risk using powerful new tools to simply optimize outdated ways of working. It’s the difference between trying to make a horse go faster and choosing to use a car instead.
The second challenge relates to how AI will be used by the workforce. Experienced shipbuilders may struggle to adopt an unfamiliar tool, while less-experienced workers may be tempted to over-rely on AI without fully understanding its limitations. Both scenarios create risk if AI is introduced without the right balance of trust, oversight, and domain expertise.
Our Data Was Built for a Different Era
AI thrives on data — but not just any data. It needs structured, contextual and accessible information to deliver meaningful insights. That’s where shipbuilding may run aground with some of the problems we want AI to solve.
For decades, our industry operated in a drawing-centric world. Even as we adopted intelligent 3D modeling tools, we continued to rely on 2D documentation to drive planning, purchasing, production and pretty much every other process. ERP systems were often fed by hand, and in some cases it was not uncommon for bills of material to be manually extracted from 2D drawings.
We have transitioned to intelligent 3D modeling, but it is only relatively recent that we are capturing the full value of these intelligent ship models to drive downstream systems.
The result? A mountain of information, but not the kind AI can easily digest.
Only recently have shipyards begun to fully leverage intelligent 3D models to feed downstream systems such as MRP, ERP and MES, mostly using an information platform such as PLM. This shift is more than a technical upgrade — it’s a strategic enabler. It allows us to manage change faster, more accurately and with less disruption.
This also highlights a gap. Much of our historical data reflects how we used to work, not how we need to work going forward. The new data we are now storing contains the digital knowledge that can be leveraged for future AI models and solutions — and that’s where real improvement will come from.
If we train AI on legacy workflows and data, we risk reinforcing outdated practices. That’s why it’s critical to focus on areas where data is already digital, structured and aligned with today’s shipbuilding methodologies. As mentioned in Part One of this series, planning, scheduling and change management are ripe with information we can leverage. The same is true for class society activities and MRO processes, such as digital images of areas of concern.
AI can still extract value from older documentation, but its real potential lies in helping us move forward — not just optimizing the past.
A Bi-modal Workforce in Transition
The second challenge is our current experienced workforce and a new, less experienced but more digitally savvy workforce. Shipbuilding is seeing a wave of retirements, and with it, a loss of deep, hard-earned knowledge. This isn’t an industry you learn in six months. The complexity lies not just in the ships we build, but in the orchestration of processes required to build them.
At the same time, the next generation entering the workforce is tech-savvy and eager to embrace AI. That’s encouraging, but it also creates tension. Experienced shipbuilders may be hesitant to trust AI, while newer team members may over-rely on it.
We need both groups to engage critically. AI should be treated like a seasoned expert — valuable and insightful, but not infallible. It can accelerate learning and improve decision-making, but only when guided by people who understand the nuances of the business.
That’s why we need to encourage teams to use AI as a tool, not a crutch. The goal isn’t to replicate past decisions, but to understand them — and then challenge them when better options exist.
The Path Forward
These two challenges — data and workforce — are deeply connected. If AI is trained on outdated data, it may reinforce old habits. If the workforce lacks context or critical thinking, it may accept AI outputs without question. In both cases, we miss the opportunity to evolve.
Real transformation happens when experienced shipbuilders use AI to challenge legacy thinking, and when newer team members use AI to learn faster while still asking the right questions. That’s the feedback loop we need: human insight guiding machine intelligence, and vice versa.
Closing Remarks
AI won’t transform shipbuilding on its own. But paired with the right data and the right mindset, it can help us build better ships, faster and with fewer surprises along the way. We have the opportunity to move fast — but we must also move forward in a smart way.
As we accelerate into a future with AI in shipbuilding, the opportunity is real, but so is the responsibility. We need to understand the data that trains our AI models to ensure our bi-modal workforce is empowered to use AI with both curiosity and discernment. The goal isn’t to digitize tradition; it’s to evolve it. If we get this right, AI won’t just accelerate shipbuilding — it will elevate it.
Let’s make sure we’re building the future with intention, not just momentum.