DNV Starts Teaching Maritime AI — and Names the Blockers Out Loud

DNV Maritime Academy has built a set of AI courses for maritime professionals, and in a Q&A published on 1 August, DNV Maritime Advisory's Head of Maritime Technology & Innovation, Dr Piyush Raj, described the obstacles in terms that should sound familiar to anyone who read last week's 11% figure. The courses pair AI fundamentals with maritime-specific application across operations, maintenance, safety, fleet management and decision support, and deliberately include the unglamorous layer: regulation, governance, cyber security, assurance and human factors. The stated aim is capability to evaluate AI — identify high-value use cases, judge vendor claims critically, understand the regulatory implications, and decide what not to adopt. In Singapore, Maritime Cluster Fund training grants are available to eligible participants through the MPA. The four blockers Raj names: data quality — organisations hold large datasets but cannot standardise them; governance — cyber security and control-and-assurance problems affecting both the data and the systems; scaling — moving past the pilot requires organisational buy-in and process integration, not better models; and trust — maritime must demonstrate AI meets the same safety bar as the technologies it replaces. His framing of the last one is the sentence worth pinning up: "Safety is the backbone of how shipping has created the modern global economy. If we lose confidence… we've lost everything." Why this matters for ship managers: last week's research found 81% of maritime companies piloting AI and 11% with policies to scale it. That gap is a skills and governance gap, and it does not close by buying another platform. When a class society starts teaching AI evaluation rather than AI enthusiasm, the market is admitting that the scarce resource is people who can tell a good system from a good demo. What to do now: identify who in your organisation is allowed to say no to an AI vendor, and check whether that person has been trained to. If the answer is your superintendent's instinct, you have a governance gap regardless of how many pilots are running. The honest caveat: this is a class society selling training, and its commercial interest in maritime AI assurance is direct. That does not make the diagnosis wrong — it matches independent research — but it does mean the course is a product, and there is as yet no published curriculum benchmark, certification standard or outcome data to judge it against. [Manifold Times]

This story is part of the Maritime AI Digest of 2 August 2026.