When the maritime trade union Nautilus International asked memberswhat they thought of AI at a forum in January, there was some positive sentiment:
“We shouldn’t automatically assume there will be problems with AI, as we’ve adopted other tech over the years without it being too intrusive for crew.”
Others were not so sure:
“There are worrying issues around liability and culpability if an incident happens involving a vessel operated using AI. Seafarers could take the blame when the fault was with the tech or the company using it.”
A lot of AI development has been focused on autonomous vessel and subsea vehicle operation, but it’s not just seafarers who are concerned. Dr David King Boison, AI Consultant and Senior Fellow of the Centre for International Maritime Affairs, Ghana, points to the potential harm embedded in using AI for logistics optimization.
Writing earlier this year, he points to the positive outcomes for ports such as Rotterdam and Singapore, both of which use AI-based port management technologies, but notes how congestion prediction algorithms optimized for these ports could perform poorly for African ports. “These systems may be trained on data from hyper-digitized ports, making their predictive models ill-suited to ports that deal with electricity outages, paper-based workflows or manual cargo inspection.”
He cites the port of Mombasa, Kenya, which piloted a predictive congestion monitoring tool in 2022. The system, based on algorithms calibrated using European port data, repeatedly underestimated truck turnaround times due to unmodeled variables such as informal gate queuing and border-related delays. This led to significant planning inefficiencies and shipment delays.
Similarly, the port of Lagos, Nigeria, has faced frequent disruptions due to AI-based customs processing tools that misclassify goods. These classification errors, often linked to language inconsistencies and poorly labeled datasets, have caused extended clearance delays, affecting exporters and importers alike, says Boison.
Biased data inputs can institutionalize discrimination even when decision-makers believe the system to be neutral, he says. Maritime examples could include:
• Cargo routing algorithms deprioritizing ports in West Africa due to perceived delays.
• Inspection algorithms flagging vessels from lower-income countries at higher rates due to biased historical datasets.
• Emissions scoring tools penalizing older ships disproportionately without considering fleet replacement limitations in developing economies.
Black-box AI systems, particularly those deployed in safety-critical sectors, can undermine trust and accountability due to the lack of explainability, says Boison. This concern is compounded in maritime contexts where jurisdictional overlaps and complex international legal frameworks already challenge accident investigations, compliance assessments and claims resolution.
The development of AI accountability, explainable AI (XAI), offers some transparency. As researchers in a recent paper in Ocean Engineering explain: For AI to be trusted and deployed in safety-critical maritime operations (e.g. collision avoidance, fault detection and autonomous vessel routing), it must provide interpretable outputs. Explainable AI techniques help uncover the rationale behind model decisions, enhancing transparency, user confidence and regulatory compliance.
These researchers, from Osaka Metropolitan University, have developed an explainable AI model for ships that quantifies the collision risk for all vessels in a given area. Graduate student Hitoshi Yoshioka and Professor Hirotada Hashimoto created the AI model which explains the basis for its decisions and the intention behind actions using numerical values for collision risk.
“By being able to explain the basis for the judgments and behavioral intentions of AI-based autonomous ship navigation, I think we can earn the trust of maritime workers,” Hashimoto stated.
The IMO is also working on the challenges posed by the uptake of digitalization and AI. During its 49th session in March, the Facilitation Committee (FAL) outlined a work plan for developing the IMO Strategy on Maritime Digitalization which is set to be adopted by the IMO Assembly by the end of 2027.
At the time, IMO Secretary-General Arsenio Dominguez emphasized the transformative potential of cutting-edge technologies such as AI and autonomous navigation, while recognizing related challenges, including cybersecurity risks and the global digital divide. "The IMO Maritime Digitalization Strategy is a game-changing effort to make smooth, seamless, smart shipping a reality. It will help integrate vessels and ports, improve logistics and optimize routes, while reducing greenhouse gas emissions. We must work together to ensure the strategy serves all.”