The factors driving marine autonomy are essentially the same as those in other transportation modes: increased safety and reduced costs. Autonomy is also touted as a means to monitor engine performance with fossil and alternative fuels (support decarbonization efforts), improve route optimization (lower fuel consumption), and aid in collision avoidance (prevent catastrophic losses). Similar to driverless trucking rationale, some proponents suggest that there will be a future workforce shortage because younger generations are not willing to go to sea. Some shipping companies are already sending up signal flares that they are having trouble finding crews due to COVID-19, which has forced crews to spend months at sea in relative isolation. Crew shortage complications could, in turn, further threaten cargo backlogs.
Are We There Yet?
To be clear, the shipping industry is not poised to go to fully uncrewed operations overnight – or anytime soon. The suite of options being explored for use on vessels ranges from a fully crewed vessel with advanced autonomy to aid in decision-making and monitor onboard efficiencies, to remote operations with crewed or uncrewed vessels, to fully uncrewed vessels with the ability to make independent operational decisions. In other words, “autonomy” does not mean “uncrewed.” Current commercial efforts focus on so-called “dull, dirty, dangerous” operations in the marine space. Testing is being done in a modern-day version of the “space race” as companies vie to be first in many feats, be it trans-Atlantic voyages, use of unmanned vessels to recover rocket boosters, transiting the Panama Canal, or remote operations from across the ocean.
Tangentially, the U.S. Navy, National Oceanic and Atmospheric Administration (NOAA), and U.S. Coast Guard have all increased research and development for marine autonomy as they see their respective future fleets embracing advanced autonomy. Notably, since both commercial and government vessels operate in shared waterways under the same “rules of the road,” the implementation of “autonomous vessels” is shifting the risk landscape, raising questions as to liability and generally challenging the status quo of a historically conservative industry.
The evolution of advanced autonomy is not revolutionary but rather an evolution in innovation, the full potential of which is yet to be determined. Innovations such as moving from sail to steam, from hand steering to autopilot, and from manned engine rooms to unmanned operations have each required both a technical and cultural shift, though it can be argued they have brought with them increased efficiencies. However, what is different with advanced autonomy is that previously, the overarching regulatory and legal regimes have been fundamentally enshrined in the concept that a human is both onboard and in charge when operating the vessel. The human element has allowed for a liability regime where responsibility and fault can be traced to a human individual (in many cases, the ship’s captain or the shipowner) when there is an incident. The human element has also formed the basis for insurance underwriting and coverage.
Liability Goes to Sea
Generally, legal liability requires a party to be responsible for the outcomes of its actions. This concept is of particular interest in the marine sector, which has historically been based on not only human interaction but a human in charge and responsible for negligent operations. However, how will liability and its consequences be assessed when incidents involve operations involving machine learning, AI programmers, code developers, or technology providers? Can the machine be at fault? Can AI have its own legal personality, and can AI be negligent? To what extent can AI fill the traditional role of the shipmaster?
Suppose humans are removed from the onboard equation, and the machine is responsible for the decision-making. In that case, this undoubtedly will lead to cracks in the legal foundation and strain current rules for vessel operations, legal precedent related to marine incidents, contractual obligations, terms in standard industry contracts, and liability regimes (including products liability and potential strict liability applications). Should a crewless vessel be involved in marine casualty, the owner may attempt to limit liability if it can prove it did not possess “privity or knowledge” of the negligent acts or unseaworthy conditions that caused or contributed to the accident. Thus, in litigation following such incidents, we may also see heightened analysis of how a shipowner can prove the absence of privity or knowledge of negligent acts. For example, is it enough to ensure that a vessel’s computer antivirus protections were updated? Will new experts be needed to assess the examination of an AI system to consider whether a vessel was seaworthy? Thus far, we have limited maritime cases on which to rely for guidance, though case law involving assessments of an onboard “glitch” offers signals of how courts may view such situations. Much remains in the realm of the unknown.
Increased automation could also lead to greater scrutiny of product liability aboard ships. Product liability law is part of the general maritime law under which the manufacturer and other entities in the chain of distribution can be held liable for harm caused by the ship. A manufacturer can be held liable for manufacturing or selling an unreasonably dangerous or defective product, which could be based on negligent or defective design, and could be measured against situations in which the risks inherent in the product outweigh its potential utility, or there was a negligent failure to warn of known defects. With more machine involvement and decision-making, the marine sector could increase interest in product liability’s role with a reduced crew or uncrewed operations.
Conclusions
Although several questions remain surrounding insurance, commercial operations, and legal liability in AI, what is clear is that counsel will play a vital role in defining the future legal regimes in current and future regulatory interpretations, litigating AI-related incidents, and aiding the insurance market in loss prevention and bespoke terms of coverage. Although the vessel crew may be at risk of being removed from the ship, the role of counsel to assess emerging risks is concomitantly only growing.
