A Reflection on How to Perform Seaworthiness Obligation in Autonomous Vessels from a Legal Perspective

In maritime law, the duty of the carrier to make the vessel seaworthy is one of the fundamental obligations in the contracts of carriage of goods by sea, so that the non-performance of this obligation may lead to the contractual liability of the carrier towards the owner of the goods. The aforementioned obligation is considered to be a multifaceted obligation, one of the important aspects of which is the hiring of a competent crew to sail the vessel to its intended destination. Therefore, if there is no crew on board or the crew recruited does not have the necessary qualifications, there is no doubt that the vessel is unseaworthy.
The recent emergence and expansion of artificial intelligence has impacted nearly all industries, including shipping. Large companies in the shipping industry are currently focused on constructing autonomous vessels using this technology. The vessels are constructed with the main approach of relying solely on artificial intelligence for navigation at sea, without the intervention of a crew. This replacement of the crew with artificial intelligence raises fundamental legal questions. One such question is how the carrier can fulfill its seaworthiness obligation without the presence of a crew. In this regard, it is important to consider whether existing international rules can be applied to autonomous vessels or if new rules should be created to account for the unique nature of this vessel class.
From a technical perspective, autonomous vessels can have varying degrees of autonomy. However, current research primarily focuses on fully-autonomous vessels. The full degree of autonomy in this category of vessels means that the vessel is able to make decisions and determine the necessary actions without the intervention of humans and solely by relying on its operating system. In response to the above-mentioned questions, different approaches have been proposed. While emphasizing the effectiveness of existing traditional rules, some believe that based on the concept of “functional equivalence,” existing international rules can also be applied to this category of vessels. In fact, according to this approach, artificial intelligence in fully autonomous vessels is supposed to do the same work that crews do in conventional vessels; therefore, based on the functional equivalence, the existing rules can be applied to artificial intelligence. On the other side, another group, referring to the “black box” problem of artificial intelligence, believes that the reformulation of the doctrine of seaworthiness should be put on the agenda in such a way that the seaworthiness of the vessel can be assessed by evaluating the algorithms used in artificial intelligence. The said problem can be explained in this way: although the input data and the output result of artificial intelligence are comprehensible by humans, humans cannot understand the decision process taken by artificial intelligence and its reason. There is, therefore, no way to evaluate whether the algorithms used in artificial intelligence are suitable for the intended sea voyage or not, and inevitably, the concept of seaworthiness must be reformulated. Finally, the third approach finds the solution in revising the existing rules because the nature of artificial intelligence requires that appropriate rules be imposed to assess the seaworthiness of autonomous vessels, and the existing rules cannot be interpreted in such a way to be applied to the mentioned vessels.