The world’s fish stocks are in decline and our increasing demand for seafood could also be one among the fundamental drivers. But the true extent of the issue is tough to estimate, especially when fishing occurs within the high seas, which lie beyond national jurisdiction and are hard to watch.

Conservation planners face growing pressures to combat illegal, unregulated and unreported (IUU) fishing, the worth of which has been estimated at US$10-23.5 billion annually. This is a very important cost for society as an entire, but in addition for the most important high seas fishing countries resembling China and Taiwan that subsidize their fleets and can have low labour costs.

High-seas vessels by flag state and equipment type, as detected by Global Fishing Watch in 2016.
From ‘The economics of fishing the high seas,’ Science Advances, CC BY-NC

Artificial intelligence (AI) could address this global environmental concern — and satisfy the necessity of seafood retailers and consumers to know if what they’re selling and eating is sustainable. Social scientists are starting to think about ways in which can bring AI, ecology and economics together — to design policies that concentrate on socially desirable outcomes resembling preserving biodiversity values and returning the advantages of fishing to society.

At a February meeting of HUMAINT, a European Commission-led initiative on human behaviour and machine intelligence, I discussed the ways AI could be used to assist marine resource management.

Poached fish

Fisheries and conservation managers have put loads of effort in recent times in establishing spatial management tools resembling marine protected areas to assist fish stocks recuperate from past over-exploitation. Fish biomass in no-take marine reserves could be on average 670 per cent greater than in unprotected areas.

Even though they’re protected, these areas are usually not at all times proof against IUU fishing. Poaching occurs and can’t be tracked easily. This could make it difficult to judge the effectiveness of the protected area in a rigorous scientific manner.

A tuna fishing port in Japan.

IUU fishing ends in environmental, economic and social costs — namely declining fish stocks — and may result in a lack of profit for those fishers who play by the foundations. It can turn the industry against the regulatory authorities that impose these spatial restrictions, undermine public trust in fisheries management and conservation science.

Tracking fishing with AI

Traditionally, observers have been employed, at high cost, to watch fishing activity on board vessels. But in distant locations, resembling the Arctic, it might be difficult to search out observers.

AI tools have the potential to lower monitoring and operational fishing costs and improve efficiency in fisheries management. Examples include automatic review of video footage, monitoring vessel sailing patterns for IUU fishing and illegal at-sea transshipments (moving goods from one ship to a different), compliance with catch limits and bycatch or discard regulations, and improving assessment of fish stocks.

AI tools can even help construct trust amongst fishers, scientists and society through improved seafood traceability.

Snow Crab is an invasive species that has been fished commercially in international waters of the Barents Sea and the Svalbard Fisheries Protection Zone of the Arctic.

Image recognition using AI may help discover the scale of a vessel and its activity. It may help conservation managers understand who fishes for what in international waters where it’s unclear who the fish belong to. It may contribute to a greater understanding of how commercially fished invasive species are spreading.

However, there are also potential risks. Some fear the information could also be used for unintended purposes or that AI tools might replace manually performed tasks and make human labour obsolete, a giant concern for small, coastal fishery-dependent communities.

The way forward

The Global Fishing Watch platform, an independent organization that emerged through a collaboration between Google, SkyTruth (a digital mapping non-profit organization) and Oceana, is a superb example of how combining AI and satellite data can change our understanding of world fishing activity.

Global Fishing Watch shows vessel movement in near real-time. Its work goes beyond tracking vessel activity: the neural network (computer program) it uses can discover vessel size and engine power, the style of fishing being done and the gear used. The ambitious project goes so far as tracking human slavery and rights abuse, a widely known phenomenon within the fishing industry.

The developments in AI applications have been impressive in recent times, allowing for a greater understanding of fishing activity across the globe. Further progress in making them more widely applicable has been limited partly by the prices involved for the industry. Concerns in regards to the impact of digital surveillance on privacy interests are also a problem.

Despite all of the progress in AI science and the event of advanced algorithms that improve the standard and speed of knowledge transmitted for ongoing fishing activities at sea, there continues to be little or no formalized integration of science, regulatory authorities and the fishing industry.

Making the most effective use of what AI tools must offer requires experts to transcend their disciplinary boundaries and actively collaborate — so that they can provide value to ongoing management efforts to conserve biodiversity and construct trust amongst seafood consumers.

This article was originally published at