Executive summary
This study reviews the main applications of Artificial Intelligence (AI) systems in fisheries and identifies current challenges for fisheries that have the potential to be dealt with through AI.
Legal analysis of relevant EU fisheries legislation that enables the use of AI systems
The AI-related legal analysis in fisheries found that: 1) there is no explicit reference to AI systems in the most relevant EU fisheries legislation, but there are references to digitalisation that could include AI systems; 2) the most relevant fisheries legislation is drafted in a way that enable the use of AI systems; 3) the broad-ranging nature of the Artificial Intelligence Act (AIA) proposal makes its application to the fisheries sector straight forward; 5) there are some concerns that the General Data Protection Regulation would require adaptation to the new realities brought by AI technologies; and, 6) several third countries are considering AI methods for further automation of Electronic Monitoring systems.
Analysis of the current and potential use of AI techniques in the fisheries sector
The review of AI techniques in fisheries revealed that: 1) Machine Learning (ML) approaches have been used to automate biological sample processing; 2) ML has been applied after image analysis and on acoustic data to count and measure organisms; 3) research on catch classification by species and sizes using AI has increased; 4) ML is being applied to automatically classify or determine fishers’ behaviour; 5) knowledge-based and expert systems have been applied to early warning systems and marine spatial planning; 6) traditional rule-based expert systems have been mainly applied in data-limited situations; 7) statistical approaches, Bayesian estimation, search and optimization methods are not traditionally considered AI, but can be integrated into AI systems; 8) some of the uses identified are applied to stock assessments and species distribution models; and, 9) fishing vessels could improve energy efficiency and reduce CO2 footprint by using AI systems.
Specific fisheries topics discussion on the use of AI systems
Firstly, seafood in all of the supply chain is analysed: 1) AI shows promising uses for traceability and seafood product integrity; 2) limited data generation and collection is the main barrier; 3) the processing industry is starting to use AI systems in automation processes; 4) AI proofs-of-concept have been developed into the logistics; and, 5) ML has been used to infer consumer behaviour and economic growth forecasting.
The second topic analysed is the use of AI for more selective fishing techniques: 1) there is a mismatch between fisheries selectivity improvement management goals and industry’s tools to comply with regulations; 2) species selectivity can be further improved with AI; and, 3) AI systems aimed at automated species forecasting and detection, identification and sizing of catches could allow improving fishing decisions and enable quota tracking.
The third discussion topic assesses the use of AI as a driving force for young people to seek jobs in fisheries: 1) AI, similarly to digitalisation, is likely to create new skilled jobs while decreasing the need for low skilled ones in the fisheries sector; 2) the marine transport sector has developed AI systems focused on ship failure prediction and anomaly detection that could be applied in fishing vessels; and, 3) a more digitalised and AI-based fisheries sector might attract new young talent, but will be competing against other industries currently offering higher incentives.
Good practices in fisheries that could be useful for future good practices within fisheries sector when developing or using AI are also analysed. Best practices guidelines for different fishery facets are commonly used by management organisations to increase their sustainability and AI technology should follow this example.
Finally, there are general AI groups and networks at the European level, but they lack marine domain knowledge to develop fit-for-purpose AI systems for fisheries. There is at least one European working group focusing on AI for fisheries and several fisheries groups where AI has been discussed, but there is a shortage of sufficient resources.
Conclusions and recommendations for AI systems use in fisheries
The last chapter summarizes opportunities and obstacles to the application of AI in the fisheries sector based on the findings of previous chapters.
Main opportunities identified are: 1) increased transparency of fishing activity and reduced impact on the environment, thereby improving the public image of the sector; 2) early warning, forecasting and spatial planning systems can help in the planning activities considering trade-offs between them; 3) accelerated and increased data acquisition and coverage for stock assessments, sustainability indicators evaluation and other management data needs; 4) increased economic sustainability of the fishing industry, by reducing operational costs; and, 5) the modernisation of fisheries and its subsequent attractiveness to the younger population.
Main obstacles identified are: 1) industry trust and reluctance; 2) initial costs and lack of expertise; and, 3) legal and bureaucratic uncertainty.
Although some AI approaches are considered black boxes (e.g. Artificial Neural Networks), there are other suitable AI methods to understand the basis, processes and model forecasts and their uncertainty (e.g. Bayesian Networks).
Finally, the study ends with the following policy recommendations for the best use of AI in fisheries and its supply chain:
- Amend Regulations that are or will be subject to revision in this field to include a reference to AI systems and AIA definition in paragraphs where digital transformation and new technologies are mentioned.
- Amend the AIA proposal to include fisheries sector. Its Recital 3 currently reads “[…] in healthcare, farming, education […]”, it could be amended to “[…] in healthcare, farming and fishery, education […]”.
- Promote formation of interdisciplinary fisheries experts with AI related skills and multidisciplinary teams (e.g., AI, biological, economic, and legal disciplines).
- Find ways to incentivise job opportunities and promotion of multidisciplinary and interdisciplinary experts not only in academia but also in the private fishery sector.
- Attract young workers and empower women with AI skills in fisheries sector jobs through dissemination of information programs and by providing adequate incentives.
- Promote private data collection and sharing, including appropriate data aggregation and anonymization safety protocols to facilitate industry trust.
- Support the development of good AI practices and standards for statistical validation and ground truth verification to increase consumer and industry trust in AI performance, also supported by strong science fit-for-purpose applications aligned with sustainability goals.
- Regulate the role of AI technological providers, ensuring some degree of experience in fisheries to prevent untrustworthy and not-fit-for-purpose AI systems (e.g., establishment of audited registration programs).
- Create regulations limiting the access of certain kinds of AI systems to the fisheries sector to help avoid their application in illegal or unethical activities (e.g., through regional fisheries management organisations (RFMOs) or registers for vessel compliance with sustainability practices from trustworthy organisations).
- Support the development of good AI practice guidelines in fisheries through collaboration with stakeholders and organisations (e.g., RFMOs, certification agencies, NGOs) using mechanisms and principles proven to be successful in other types of fisheries best practices guidelines.
- Promote AI awareness, both benefits and constraints, among managers and industry to improve adoption processes at the whole supply chain.
- Promote collaboration between universities, firms, AI developers and other stakeholders in fisheries though specific funding, specialized centres, and multidisciplinary networks.
- Promote technological development integrated with AI systems to develop more selective fishing gears and fishing strategies by funding AI research and vessels digitalisation.
Link to the full study: https://bit.ly/699-643
Please give us your feedback on this publication
Selection of visuals:






-
- Digitalisation refers to enabling or improving processes by using digital technologies and data. However, a digital transformation process does not always necessarily imply the use of AI, despite its increasing presence in digitalized processes
-
- Figure 2 illustrates the classification of AI techniques and approaches in the AIA proposal expanded with further subcategories used in this study.
-
- Expert systems are knowledge-based systems (Figure 3) that in an interactive setting ask a person for information and draw conclusions based upon the response, or give advice, emulating the decision-making ability of a human expert.
-
- The value chain (Figure 4) describes the full range of activities required to bring a product or service from conception, through the different phases of production (involving a combination of physical transformation and the input of various producer services), and delivery to final consumers: ‘from sea to fork’ (Rosales et al., 2017). Tracking of information in the fish food supply chain is difficult but needed for consumer trust in compliance with high levels of food safety that European legislation aims to achieve (Nicolae et al., 2017).
-
- The basic concept of traceability only includes the record and track of necessary information in all value chain steps from procurement of raw materials and parts to machining, assembly, distribution, and sales. However, advances in the data science allow more key aspects to be included. For example, the use of AI in traceability covers elements like food integrity, sustainability, processing, logistics and market
-
- The evolution of the papers published in the topic of traceability in combination with fisheries and/or AI keywords show a constant increase (Figure 6). However, only 100 publications out of 1,600 refer to traceability, and this number drops to 6 if the AI keyword is added.
2 Comments
[AT A GLANCE] Artificial Intelligence and the fisheries sector – Research4Committees · May 30, 2022 at 1:46 pm
[…] to the full study: Artificial Intelligence and the fisheries sector Loading… Taking too […]
[Digest] Study presentation – Artificial Intelligence and the fisheries sector – Research4Committees · June 14, 2022 at 3:37 pm
[…] Further reading: Artificial Intelligence and the fisheries sector […]