This insight examines how poor data foundations, fragmented systems, and cognitive overload prevent AI from delivering reliable voyage optimisation and undermine human confidence.
Many small and medium maritime companies still lack the digital foundations required for AI to work effectively. According to a CIO at a shipping company we interviewed, data is often siloed, inconsistent, or locked inside spreadsheets, which means AI models are frequently trained on fragmented or low-quality information.
This may lead to voyage optimisation systems suggesting uneconomical routes, under- or over-estimating fuel consumption, or recommending unsafe operating parameters because the underlying data did not reflect actual vessel conditions.
A major gap is the industry’s limited access to high-frequency data. While noon reports offer a partial snapshot, they are not sufficient for the continuous input required by modern optimisation tools. Outdated sensors and ageing onboard equipment delay the rollout of continuous data streams, and in some cases, poor-quality data is transmitted without anyone realising the negative impact until recommendations go wrong.
Without strong data governance and without humans continually validating, interpreting, and improving the inputs, AI cannot reach the level of accuracy required for meaningful optimisation.
At the same time, crews and office teams face an overwhelming number of dashboards, alerts, and disconnected interfaces. Information overload is becoming a growing risk to safe operations, as it increases cognitive load and contributes to human error.
One Voyage Manager told Thetius that they have seen operators get “30–40 alerts per voyage, but only five of them apply.” The question becomes whether time is wasted checking all alerts or whether they are ignored altogether.
Even when the right data is presented, people interpret it differently. Two people might look at the same weather map, dashboard, or AI routing output and see completely different things. A non-mariner might notice vessel positions and overlay colours, whereas an experienced navigator sees pressure gradients, storm evolution, or vessel stability implications.
Fragmentation between ship and shore further deepens disagreement about which AI recommendations to trust or act on, making it harder to balance human intelligence with machine intelligence.
To explore how a balanced human-AI approach can strengthen voyage optimisation in safety-critical conditions, read the full Thought Leadership report ‘Co-Pilots of the Sea: Exploring the Human Intelligence Behind Maritime AI’ produced in partnership with StormGeo.
Â
