AI can be harnessed to transform shipping by delivering leaner supply chains, lower emissions, safer fleets. While using LLMs like Claude, ChatGPT, Grok, DeepSeek etc, can help with the optimisation of personal output (such as email writing, document creation, etc), wider use of machine learning and agentic AI needs specific conditions that must be in place to deliver the results being flaunted as a game changer. 

There are specific reasons that the maritime sector’s approach to AI will be different than that of other sectors, with three structural divides shaping the conversation the most: asset lifecycles, operational liability, and hierarchy.

Asset lifecycles

Maritime leaders are often seen as luddites for not investing in emerging digital technology, but there is an extremely obvious reason at the heart of this approach. A commercial vessel runs for about 25 years, with one major retrofit planned in, meaning that maritime capital is largely expected to run for decades without major upheaval. Software isn’t. Product cycles in tech run two years at best, three to six months at worst. Models get deprecated, code gets rewritten, startups disappear. AI in particular is at the speedy end of this cycle, with new launches coming out regularly enough for something premium to be seen as run of the mill a mere 6 months later.

A maritime leader choosing what AI will run in their offices, on vessels and as the heart of their digital operations must contend with the fact that they could be committing millions to a software stack with no idea of what “good” looks like in ten years, or whether the vendor will even exist by then. Caution is not just an approach but a strategy that will be a determining factor, especially given the international nature of shipping and the increasing limitations to digital information. 

While big players are likely to pick up a lot of maritime clients as they are likely to be around in 20 years, start-ups and tech providers will struggle to get buy-in unless they are able to demonstrate enough of a return on investment to justify taking a risk on a small player.

Operational liability

Onshore, adopting new software is relatively low-stakes. In a best-case scenario, when something breaks, IT rolls it back, life goes on. In a worst case scenario, people are working hard to manage the damage, but there is no threat to life in the same way as it is at sea.

Onboard a ship, the captain carries the liability of the outcomes of an incident, irrespective of whether it was decided by a person or software. A bad maneuver or equipment failure isn’t just a bad quarter, it means harm to the ship, the crew, the environment and the cargo. Most importantly, it can also mean criminal charges.

When looking at whether the crew will embrace AI, we must factor in that (at present), the software or its creators carry zero legal consequence for getting it wrong. The captain carries all of the risk and the consequences. And if the system is a black box with no visibility into the data, no way to check the sources, no explanation of the reasoning, the captain is being asked to sign off on something they can’t verify. That’s not a risk any individual should have to take. AI at sea needs to be explainable and auditable if there is to be meaningful buy in.

Hierarchy and quality

Maritime careers are fundamentally hierarchical. Cadets earn their way to Chief Engineer or Master over years at sea, and boardrooms lean on people with decades of industry knowledge behind them. As an emerging technology, AI doesn’t conform to that ladder. Unless it has been a sector that they are heavily involved in, the people best at getting value out of these tools tend to be digital natives, who are often younger and thus likely to be junior. But that same junior officer often lacks the seasoning to catch it when the AI is subtly wrong. So you end up with a split: the people who understand the tech don’t have the experience to check its output, and the people with the experience don’t trust the tech enough to engage with it.

The same problem shows up in the boardroom. Executives are under pressure to invest in AI, but many boards don’t understand the importance of the data architecture underneath it. Fragmented, poorly standardised legacy data means garbage in, garbage out — no matter how sophisticated the AI wrapper is. Without real data literacy at the top, boards risk paying a premium for very expensive snake oil. In many cases, these individuals may not even know the state of their organisation’s data, meaning that they cannot accurately assess if the AI being invested in can spin straw into gold (as per the Rumplestiltskin fairytale) or whether they will fall victims to ‘garbage in, garbage out”.

And we must acknowledge that it is extremely difficult for junior members of the organisation to push back against a multi-million dollar investment, where the argument fundamentally involves pointing to a lack of knowledge in the senior members of the team. 

The bottom line

None of this rules AI out — it just means the industry can’t import tech’s “move fast and break things” playbook wholesale. In shipping, breaking things means lives, vessels, and ecosystems, not just a bad app update.

Getting this right means respecting 25-year asset cycles, building systems that protect — not expose — the captain’s liability, and pairing digital fluency with hard-earned sector experience rather than letting the two sit in separate rooms. Solve for those three divides, and AI stops being a risk to manage and starts being the advantage it’s supposed to be.

Words by Intent Communications CEO and Founder Namrata Nadkarni. Photo by Ian Simmonds on Unsplash
———————-
Like this article? Click here to sign up for our free monthly ‘De-Mystifying Digitalisation and AI in Maritime’ newsletter and get the latest cross-sector expert insights, news, interviews, videos and case studies straight to your inbox.