Artificial intelligence in aquaculture: human-centered innovation, ethical governance, and data foundations for sustainable blue growth

Abstract

Artificial intelligence (AI) is transforming aquaculture by enabling precision management, environmental monitoring, and sustainability-oriented decision support. This review advances the discourse by integrating human-centered AI, ethical governance, and sustainability frameworks into a cohesive analysis of digital transformation in aquaculture. Based on a structured synthesis of 220 peer-reviewed publications from multidisciplinary literature published between 2015 and 2025, the study employs a qualitative review methodology to identify emerging trends, challenges, and research directions in AI-enabled aquaculture systems. The analysis reveals three emergent research pillars: (1) human-centered and explainable AI (XAI) systems that enhance decision transparency and farmer engagement; (2) ethical and governance frameworks addressing data ownership, algorithmic bias, and accountability; and (3) technological applications and innovation pathways linking machine learning, computer vision, and Internet of Things (IoT) platforms to operational sustainability. Findings indicate that while AI-driven tools have improved biomass estimation, behavior tracking, disease detection, and feed optimization, adoption remains constrained by affordability, digital literacy, infrastructure limitations, and data interoperability barriers. Ethical concerns related to transparency, cybersecurity, privacy, and equitable access to data further underscore the need for adaptive governance mechanisms. Aligning these technological and ethical dimensions with the Food and Agriculture Organization (FAO) Blue Transformation agenda and the Organization for Economic Co-operation and Development (OECD) AI Principles highlights a pathway toward inclusive, responsible, and context-sensitive AI ecosystems in aquaculture. By bridging the technical and human dimensions of AI deployment, this synthesis proposes a conceptual framework for responsible digital aquaculture in which innovation is embedded within social, ethical, and policy-responsive systems. The review concludes that the long-term sustainability of AI in aquaculture will depend not only on technological advancement but also on the co-evolution of governance structures, human capacity, and environmental stewardship.

1 Introduction

Aquaculture has become one of the fastest-growing food production sectors globally and now surpasses capture fisheries as the primary source of aquatic foods for human consumption (FAO, 2022). As the global population is projected to reach 9.7 billion by 2050 and more than 730 million people experienced undernourishment in 2023, the sector plays an increasingly critical role in ensuring food security, nutritional adequacy, and economic development (Boyd et al., 2022; Obirikorang et al., 2024). However, the rapid expansion of aquaculture has also intensified concerns regarding environmental sustainability, resource efficiency, and social equity. Persistent dependence on fishmeal and fish oil derived from wild fisheries, climate-related vulnerabilities, disease outbreaks, and unequal access to technological innovations continue to challenge the long-term sustainability of aquaculture systems (Boyd et al., 2022).

In response to these challenges, global policy initiatives have increasingly emphasized the need for transformative approaches to aquatic food production. The Food and Agriculture Organization’s (FAO) Blue Transformation agenda promotes the development of sustainable, resilient, and inclusive aquatic food systems through innovation, responsible resource management, and the empowerment of small-scale producers, women, and youth (Agostini et al., 2025). Achieving these objectives requires governance systems capable of integrating ecological, social, and economic priorities while fostering collaboration across sectors and scales. Such systems must support transparency, accountability, and connectivity across land–water–sea interfaces to address the complex and interdependent challenges facing contemporary aquaculture (Partelow et al., 2023).

Digital technologies are increasingly recognized as important enablers of this transformation. The proliferation of the Internet of Things (IoT), blockchain platforms, cloud computing, and Artificial Intelligence (AI) has expanded opportunities for data-driven management, traceability, and adaptive decision-making within aquaculture operations (Rowan, 2022). Among these technologies, AI has emerged as a particularly influential innovation due to its capacity to analyze large and complex datasets, automate monitoring processes, identify patterns in environmental and biological systems, and generate predictive insights for farm management. Recent advances in machine learning, deep learning, computer vision, and generative AI have enabled applications ranging from automated feeding systems and water-quality monitoring to disease detection, biomass estimation, behavioral analysis, and production forecasting (Rather et al., 2024; Huang and Khabusi, 2025).

Despite these advances, the integration of AI into aquaculture remains fragmented and uneven. Existing studies have largely focused on technical performance and operational efficiency, while comparatively limited attention has been devoted to understanding how AI technologies interact with broader sustainability goals, governance structures, and stakeholder needs. Adoption remains constrained by barriers including high implementation costs, limited digital infrastructure, data silos, interoperability challenges, and disparities in technological access. At the same time, concerns surrounding algorithmic bias, transparency, cybersecurity, data ownership, and accountability raise important questions about the responsible deployment of AI in aquaculture systems. These challenges suggest that technological innovation alone is insufficient to achieve sustainable transformation and that greater attention must be given to the social, ethical, and institutional dimensions of AI adoption.

Consequently, there remains a significant gap in the literature regarding how AI can be integrated into governance frameworks that promote sustainability, resilience, and inclusivity across aquaculture value chains. Addressing this gap requires a comprehensive understanding of how AI is currently applied throughout aquaculture production systems, what governance and ethical challenges influence its adoption, and how emerging Human-Centered Artificial Intelligence (HCAI) principles can support more equitable and transparent forms of digital transformation. By examining these interconnected dimensions, scholars and practitioners can better understand the conditions under which AI contributes not only to operational efficiency but also to broader societal and environmental objectives.

This review addresses these issues through a synthesis of AI applications in aquaculture from 2020 to 2025. It examines the evolving roles of machine learning, deep learning, computer vision, and generative AI in supporting evidence-based decision-making and adaptive management across aquaculture systems (Hossain et al., 2024; Akram et al., 2025). In doing so, the review evaluates both the technological opportunities and the governance challenges associated with AI adoption, while exploring the potential of human-centered approaches to enhance transparency, stakeholder participation, trust, and accountability. Through this integrated perspective, the study seeks to bridge the divide between technological innovation and sustainability governance, providing a conceptual foundation for responsible and inclusive AI implementation in aquaculture.

Recent reviews further demonstrate the rapid expansion of AI research in aquaculture. Öz and Üstüner (2026) emphasized the role of predictive modeling and decision-support systems in advancing Blue Transformation and sustainable aquaculture, while Öz et al. (2025) reviewed AI innovations in fish disease diagnosis and management, including their biosecurity and One Health implications. These studies provide important foundations for understanding AI’s technical applications; however, further synthesis is needed to connect these advances with human-centered design, ethical governance, data interoperability, and inclusive implementation.

Furthermore, this review differs from recent AI-in-aquaculture reviews by moving beyond application-specific or technology-centered syntheses and by explicitly integrating human-centered AI, ethical governance, data foundations, and sustainability-oriented implementation within a unified analytical framework. While recent reviews have examined AI-enabled decision support, Blue Transformation, fish disease diagnosis, and One Health implications, the present review advances the field by linking these domains to explainability, farmer adoption, data ownership, algorithmic accountability, interoperability, and inclusive governance. Specifically, this review aims to: (1) synthesize major AI applications across aquaculture production, health, monitoring, and supply-chain systems; (2) evaluate the data, infrastructure, and governance conditions required for responsible AI adoption; (3) examine how human-centered and explainable AI can improve trust, transparency, and stakeholder participation; and (4) propose an integrated AI–Sustainability–Governance Framework for responsible digital aquaculture.

Conceptually, the study positions AI as a catalyst linking technological innovation, sustainability outcomes, and governance transformation within aquaculture systems (Figure 1). The proposed AI–Sustainability–Governance Framework illustrates a progression from AI technologies to data-driven aquaculture systems, which in turn contribute to sustainability outcomes such as efficiency, resilience, and inclusivity. These outcomes are subsequently reinforced through governance integration involving policies, institutions, stakeholder participation, and ethical oversight. By connecting technological capabilities with governance and sustainability objectives, the framework offers a pathway toward realizing the Blue Transformation vision, where responsible innovation supports environmentally sustainable, socially inclusive, and economically resilient aquaculture systems.

Figure 1

2 Methodology

2.1 Structured thematic synthesis approach

This review employed a structured thematic synthesis approach to examine the evolving role of artificial intelligence (AI) in aquaculture and its implications for sustainability, governance, and digital transformation. Given the interdisciplinary nature of AI research and the rapid emergence of new technologies, a concept-driven review design was adopted to integrate evidence from technological, environmental, social, and policy perspectives. Rather than conducting a quantitative meta-analysis, the study synthesized scholarly and institutional literature to identify recurring themes, knowledge gaps, and future research directions relevant to AI-enabled aquaculture systems.

2.2 Literature curation and search strategy

An extensive literature search was conducted to identify publications addressing the intersection of AI, aquaculture, sustainability, and governance. The review focused on literature published between 2015 and 2025 to capture recent developments in digital aquaculture and emerging AI technologies. Sources were retrieved from major academic databases, including Scopus, ScienceDirect, and SpringerLink, and supplemented with reports, policy documents, and technical publications from international organizations such as the Food and Agriculture Organization (FAO), the Organisation for Economic Co-operation and Development (OECD), and the United Nations Educational, Scientific and Cultural Organization (UNESCO).

The search was conducted using combinations of Boolean keywords across the selected databases. Search strings included: (“artificial intelligence” OR “machine learning” OR “deep learning” OR “computer vision” OR “predictive analytics” OR “generative AI”) AND (“aquaculture” OR “fish farming” OR “shrimp farming” OR “mariculture” OR “recirculating aquaculture system”) AND (“sustainability” OR “governance” OR “ethics” OR “decision support” OR “biosecurity” OR “traceability” OR “Blue Transformation”). Database searches were supplemented through backward and forward citation tracking of highly relevant review articles, empirical studies, and institutional reports.

The search strategy combined keywords representing both technological and governance dimensions of AI adoption in aquaculture. Core search terms included artificial intelligence, aquaculture, machine learning, computer vision, Internet of Things (IoT), digital aquaculture, explainable AI, ethics, governance, and blue economy. Additional publications were identified through reference-list screening and citation tracking of influential review papers and policy reports.

2.3 Literature screening and selection

Retrieved documents underwent a multi-stage screening process to ensure relevance to the objectives of the review. Titles, abstracts, and full texts were examined to assess their contribution to understanding AI applications, sustainability outcomes, governance considerations, or innovation pathways within aquaculture systems.

Studies were included when they addressed one or more of the following areas:

• AI, machine learning, deep learning, computer vision, generative AI, or IoT applications in aquaculture;

• Digital technologies supporting aquaculture production, monitoring, health management, or decision-making;

• Ethical, governance, regulatory, or policy dimensions of AI deployment;

• Sustainability outcomes related to environmental performance, resource efficiency, resilience, or social inclusion.

Publications that focused exclusively on capture fisheries, lacked substantive relevance to aquaculture, or provided insufficient conceptual or technical detail were excluded from the synthesis. To improve transparency, the final set of 220 publications was selected based on four criteria: thematic relevance, methodological or conceptual contribution, source credibility, and applicability to aquaculture systems. Priority was given to peer-reviewed studies that provided empirical evidence, validated models, systematic reviews, or clearly described AI applications in aquaculture. Policy reports and institutional documents were included only when they provided authoritative governance, sustainability, or ethical frameworks relevant to AI deployment. Each publication was assessed for clarity of objectives, relevance to aquaculture, adequacy of methodological description, strength of evidence, and contribution to one or more thematic categories. Studies with unclear methods, weak relevance to aquaculture, duplicate content, or insufficient technical or conceptual detail were excluded.

Publications were included if they: (1) addressed AI, machine learning, deep learning, computer vision, IoT, digital twins, blockchain, or related digital technologies in aquaculture; (2) focused on aquaculture production, health management, environmental monitoring, supply chains, governance, ethics, or sustainability; (3) provided empirical evidence, conceptual synthesis, technical validation, or policy relevance; and (4) were published in English between 2015 and 2025. Publications were excluded if they focused only on capture fisheries, lacked clear relevance to aquaculture, provided insufficient methodological or conceptual detail, duplicated another source, or discussed AI only superficially without connection to aquaculture systems.

2.4 Data extraction and thematic categorization

Following screening, selected documents were reviewed systematically to extract information on AI technologies, application domains, governance frameworks, adoption barriers, sustainability outcomes, and future development pathways. Extracted information was organized and compared across studies to identify recurring patterns and conceptual relationships.

An iterative thematic analysis was then conducted to group findings into broader thematic categories. Through repeated review and comparison of the literature, four dominant themes emerged:

• Human-Centered AI and Farmer Adoption – focusing on usability, trust, explainability, digital literacy, and stakeholder engagement;

• Ethics, Governance, and Policy Dimensions – addressing transparency, accountability, data ownership, cybersecurity, and regulatory frameworks;

• Case Studies and Industrial Practices – synthesizing practical applications of AI technologies across aquaculture production systems;

• Research Frontiers and Future Directions (2025–2030) – identifying emerging technologies, innovation opportunities, and unresolved research challenges.

These themes provided the analytical structure for the subsequent sections of the review.

2.5 Conceptual triangulation and framework development

To strengthen interpretive rigor, the review employed conceptual triangulation by aligning empirical findings with internationally recognized sustainability and governance frameworks. In particular, the analysis was informed by the FAO Blue Transformation Agenda and the OECD AI Principles, which provide complementary perspectives on sustainable aquatic food systems and responsible AI governance.

The integration of technological evidence, thematic findings, and policy frameworks informed the development of the AI–Sustainability–Governance Framework proposed in this study. This framework conceptualizes AI as a catalyst linking technological innovation, sustainability outcomes, and governance transformation within aquaculture systems. Through this approach, the review identifies both the opportunities and challenges associated with AI adoption while highlighting pathways toward responsible, inclusive, and sustainable digital aquaculture. The overall literature curation, screening, thematic synthesis, conceptual triangulation, quality assessment, and framework-development process is summarized in Figure 2.

Figure 2

2.6 Quality assessment and reproducibility safeguards

To strengthen methodological rigor, each retained publication was assessed using five quality criteria: relevance to aquaculture, clarity of objectives, adequacy of methodological description, strength of empirical or conceptual evidence, and contribution to the review themes. Empirical studies were evaluated based on the clarity of model description, dataset source, validation approach, and reported limitations. Review papers were assessed based on scope, synthesis depth, and relevance to AI-enabled aquaculture. Institutional documents were included only when they provided authoritative guidance on governance, sustainability, ethics, food systems, or responsible AI. This appraisal process ensured that the final synthesis prioritized sources with clear relevance, methodological transparency, and conceptual contribution.

2.7 Conceptual framework development and figure preparation

The conceptual figures presented in this review were developed by the author as original graphical syntheses of the themes, mechanisms, technological architectures, governance dimensions, and conceptual relationships identified through the structured thematic synthesis of the literature. Figure development followed an iterative process involving: (1) identification of recurring concepts and relationships across the reviewed studies; (2) organization of these concepts into thematic, mechanistic, or systems-level frameworks; (3) development of preliminary visual structures representing relationships among technologies, biological processes, data flows, stakeholders, governance mechanisms, and sustainability outcomes; and (4) graphical refinement to improve conceptual coherence, scientific accuracy, visual clarity, and consistency with the narrative synthesis.

The figures were graphically designed, assembled, and refined using Canva (Canva Pty Ltd., Sydney, Australia). Graphical elements, icons, text, directional relationships, and overall layouts were selected, organized, and refined by the authors to visually represent the conceptual relationships identified through the literature synthesis. The author was responsible for the conceptual development, organization of scientific content, verification of the graphical representations against the supporting literature, and final presentation of all figures.

The figures do not present independently generated empirical data or quantitative analyses; rather, they constitute author-developed conceptual syntheses based on the scientific and institutional literature reviewed in the corresponding sections of the manuscript. The figures were developed specifically for this review and were not reproduced or adapted from previously published figures unless explicitly indicated. Relevant references supporting the principal concepts, technological processes, and thematic relationships illustrated in the figures are provided in the corresponding sections of the manuscript and figure captions.

3 Data foundations for AI in aquaculture

3.1 Sensors and modalities

AI-driven aquaculture systems increasingly depend on multimodal sensing technologies that generate the foundational data required for real-time analytics, automation, and decision support. Optical sensors, including RGB and multispectral imaging, facilitate non-invasive assessments of fish biomass, health, and water quality (Saberioon et al., 2017; Huang and Khabusi, 2025). Internet of Things (IoT) networks integrate temperature, pH, dissolved oxygen, and salinity sensors, while biosensors detect pathogens and biochemical hazards (Sharma and Kumar, 2021; Su et al., 2020). Acoustic and sonar systems complement visual data by enabling biomass estimation and behavioral monitoring in turbid waters, enhanced by deep learning architectures such as Mask R-CNN (Li et al., 2023; Chang et al., 2022). The overall integration of these multimodal data streams within AI-enabled aquaculture ecosystems is illustrated in Figure 3, which depicts the interconnections among sensor modalities, IoT networks, and cloud-based analytics workflows.

Figure 3

Emerging technologies, including digital twins, generative AI, and autonomous remotely operated vehicles (ROVs), extend the capabilities of sensing and modeling but remain constrained by environmental variability, data fragmentation, and GNSS accuracy limitations (Akram et al., 2025; Skaldebø et al., 2024). Nevertheless, integration of AI-enabled sensors has shown measurable productivity gains, with reductions in fish mortality by up to 40% and yield increases of 15–50% (Liu et al., 2025).

3.2 Data pipelines and computational architectures

The rapid expansion of aquaculture data streams necessitates scalable computational infrastructures. Edge–fog–cloud architectures now underpin real-time processing and distributed analytics, allowing low-latency decision-making at the farm level while maintaining centralized oversight (Cheng et al., 2024; Kalyani et al., 2024). Federated learning models further enhance privacy and data sovereignty by enabling decentralized training across farms without direct data sharing. Digital twin platforms combine IoT and AI models to simulate system dynamics and optimize feeding, energy use, and water quality management (Ubiña et al., 2023).

To manage complexity, Machine Learning Operations (MLOps) frameworks such as Pangea automate model deployment, monitoring, and continuous integration within heterogeneous aquaculture environments (Miñón et al., 2022). These architectures form the technical foundation for scalable, adaptive, and resilient aquaculture analytics ecosystems.

3.3 Data quality and benchmarking

Data quality remains the most persistent constraint in AI-enabled aquaculture. Labeling errors, dataset bias, and class imbalance significantly undermine model reliability and transferability across contexts. Tools such as the Dataset Nutrition Label (Holland et al., 2018) and the AQuA benchmarking platform (Goswami et al., 2023) have begun standardizing quality assessment under noisy-label conditions. For instance, phytoplankton detection studies continue to suffer from limited class representation, impeding accurate forecasting of harmful algal blooms (Guterres et al., 2023). Rigorous validation and benchmarking of disease, behavioral, environmental, and production datasets remain essential to ensure reproducibility and cross-regional comparability (Ragab et al., 2024).

3.4 Standards and interoperability

The lack of automation and interoperability across aquaculture data systems continues to delay digital transformation (Abid et al., 2019). Adoption of FAIR data principles—Findable, Accessible, Interoperable, and Reusable—has emerged as a key enabler for cross-platform integration (Wilkinson et al., 2017). AIoT frameworks now emphasize standardized communication protocols, cybersecurity safeguards, and semantic web technologies to facilitate scalable, interoperable telemetry across farms and institutions (Huang and Khabusi, 2025; Rather et al., 2024).

The maturity of data sources used in AI-enabled aquaculture varies considerably. While sensor-based systems have achieved relatively high operational readiness and widespread deployment, emerging biological data streams such as environmental DNA (eDNA) and multi-omics datasets remain constrained by standardization, integration, and governance challenges. Table 1 provides a comparative assessment of the maturity and limitations of key aquaculture data sources.

The progression from sensor-based monitoring toward genomics, eDNA, and digital twin ecosystems reflects a broader transition from descriptive data collection to predictive and adaptive intelligence. However, differences in data maturity underscore the continuing need for standardized infrastructures, benchmarking frameworks, and interoperable data governance mechanisms.

3.5 Critical analysis

While significant progress has been made, the digital readiness of aquaculture infrastructure in Asia remains uneven. High-income economies such as Japan, Singapore, and South Korea are advancing toward AI-integrated digital twins and real-time production analytics, supported by robust broadband networks and skilled labor. In contrast, developing Southeast Asian countries, notably the Philippines, Indonesia, and Vietnam, demonstrate strong potential but face structural barriers, including limited data infrastructure, inconsistent internet connectivity, and gaps in digital literacy.

Data reliability and interoperability pose cross-cutting challenges. Most farm-level data remain siloed in proprietary systems or fragmented across government and private repositories, impeding model transferability and governance integration. Furthermore, inconsistent metadata standards and weak validation frameworks reduce confidence in AI-driven insights for policy or investment decisions.

Despite these challenges, emerging opportunities are reshaping the data ecosystem. Federated learning offers privacy-preserving collaboration across farms and institutions without centralizing data, while digital twins provide predictive, real-time decision support for optimizing feeding and resource use. The convergence of generative AI and edge computing promises adaptive analytics capable of functioning even in data-sparse or connectivity-limited environments, particularly relevant for smallholder aquaculture systems.

Comparatively, developed nations lead in scaling advanced analytics due to greater access to capital, computing power, and technical expertise, whereas developing countries often depend on donor-led or pilot initiatives. Bridging this divide requires investment in open data infrastructures, capacity building, and regional data governance frameworks to ensure equitable access to AI-driven innovation.

Overall, while the technological underpinnings for AI in aquaculture are rapidly advancing, the path toward sustainable, inclusive, and interoperable data ecosystems, especially in Southeast Asia, remains a pivotal frontier for the sector’s digital transformation.

3.6 Production-system-specific considerations for AI adoption

AI adoption in aquaculture is highly system-dependent. Recirculating aquaculture systems (RAS) are generally more suitable for advanced AI deployment because they offer controlled environments, continuous sensor installation, stable power supply, and dense water-quality datasets. Pond-based systems, which dominate much of smallholder aquaculture in Asia, often face greater constraints due to variable turbidity, irregular connectivity, and limited automation infrastructure. Cage and net-pen systems benefit from computer vision, acoustic monitoring, and remotely operated platforms, but are more exposed to waves, biofouling, weather variability, and sensor maintenance challenges. Hatcheries and nurseries may benefit from AI-supported water-quality control, larval monitoring, disease detection, and feeding optimization because early life stages are highly sensitive to environmental fluctuation. Offshore and mariculture systems require more robust AI architectures that combine remote sensing, acoustic systems, edge computing, and weather-risk forecasting. Therefore, AI implementation should not be treated as a uniform technological pathway; rather, system-specific feasibility, infrastructure readiness, species biology, farm scale, and farmer capacity should guide technology selection.

4 Computer vision and multimodal perception

4.1 Biomass estimation and stocking density

Computer vision has rapidly evolved as a non-intrusive alternative to manual sampling in aquaculture. Convolutional and segmentation-based deep learning models, such as U-Net and Mask R-CNN, have demonstrated strong predictive power for biomass estimation and stocking density across multiple taxa. Studies report prediction accuracies exceeding 90% for species such as shrimp and Nile tilapia (Thai et al., 2021; Fernandes et al., 2020), while correlations for macroalgae biomass also remain high but environment-dependent (Overrein et al., 2024). Despite these achievements, standardization across datasets, lighting conditions, and species morphologies remains limited. The absence of unified image benchmarks constrains model generalization and cross-system comparability, particularly under variable turbidity and lighting conditions.

To facilitate cross-study comparison, Table 2 summarizes representative computer vision models, application domains, reported performance, and key limitations identified in the literature.

The benchmark comparison highlights a broader trend in which model performance is increasingly driven by data quality, environmental conditions, and deployment context rather than algorithmic sophistication alone. Consequently, future progress will depend on the development of standardized datasets and evaluation protocols capable of supporting reproducible cross-species comparisons.

4.2 Behavior and welfare monitoring

Vision-based behavioral analysis supports welfare monitoring by detecting deviations in feeding, swimming, and stress patterns. AI systems now quantify fish depth distribution and activity levels in response to environmental stressors such as oxygen fluctuation or high-density stocking (Burke et al., 2025). Machine learning-based anomaly detection has successfully identified indicators of distress, including erratic movement and immobility (Shreesha et al., 2023), but such systems still face accuracy degradation under turbidity and overlapping body postures (Fitzgerald et al., 2025). Standardized behavioral datasets and multimodal labeling remain critical for validating welfare metrics across species and environmental contexts.

Recent advances demonstrate that welfare monitoring increasingly benefits from multimodal integration, combining visual observations with environmental sensor data and behavioral analytics. Such approaches improve contextual interpretation of fish behavior and reduce false positives arising from isolated image-based assessments.

4.3 Counting and tracking in turbid environments

Tracking and counting under low visibility rely increasingly on multimodal perception, integrating optical and acoustic data streams. Advanced multiple object tracking (MOT) algorithms and sensor fusion approaches have improved counting precision in commercial ponds and net pens (Nishiguchi et al., 2025; Cui et al., 2024). In challenging turbidity conditions, AI models maintain approximately 90% accuracy for residual feed detection and individual tracking (Cao and Xu, 2018). These results demonstrate technological robustness but also highlight the need for calibration standards to harmonize model validation across environments and sensor types.

The increasing use of multimodal perception reflects a shift away from single-sensor systems toward integrated architectures capable of compensating for environmental uncertainty. Combining acoustic, sonar, and optical inputs has proven particularly valuable in commercial settings where visibility conditions fluctuate substantially.

4.4 Edge deployment and latency solutions

The deployment of AI vision systems in operational farms is constrained by data transmission limits and computational load. Edge–cloud frameworks and lightweight convolutional models are increasingly adopted to process high-volume video locally, reducing latency and bandwidth requirements (Pajo et al., 2023; Cheng et al., 2024). YOLO-based detection systems and hybrid multimodal architectures show promise for real-time welfare assessment but remain sensitive to lighting shifts and behavioral noise (Fitzgerald et al., 2025; Wu et al., 2025).

The convergence of edge computing and computer vision is particularly important for aquaculture systems operating in remote or connectivity-constrained environments. By enabling localized inference and decision support, edge deployment reduces reliance on continuous cloud connectivity while improving responsiveness to rapidly changing farm conditions.

4.5 Critical analysis

Overall, computer vision and multimodal AI in aquaculture are transitioning from proof-of-concept technologies (TRL 4–5) toward early commercial deployment (TRL 6–7), particularly in biomass estimation, counting, and welfare monitoring applications. As summarized in Table 2, most leading models consistently achieve high predictive performance under controlled conditions; however, performance often declines when deployed across different species, farming systems, and environmental contexts.

Regional disparities remain significant. Developed economies, including Norway and Japan, benefit from integrated digital infrastructures, large, annotated datasets, and advanced computing resources. In contrast, many Southeast Asian aquaculture systems face persistent barriers related to computational access, labeling expertise, internet connectivity, and data standardization. The principal predictive-model applications, target parameters, advantages, and supporting evidence across aquaculture control systems are summarized in Table 3.

A recurring challenge across all benchmarked systems is the lack of open, interoperable datasets and universally accepted evaluation metrics. Models trained in controlled experimental settings frequently lose accuracy under real-world variability, including changing illumination, turbidity, stocking densities, and species morphology. Consequently, future progress depends on the establishment of open multimodal datasets, cross-species benchmark protocols, and federated learning frameworks that preserve data privacy while enabling collaborative model improvement. The integration of computer vision, multimodal tracking, sensor fusion, behavioral monitoring, and edge-based deployment in AI-enabled aquaculture is synthesized in Figure 4.

Figure 4

5 Predictive modeling and control

AI-driven predictive modeling has become the analytical backbone of precision aquaculture, enabling proactive management of water quality, feeding, and biomass dynamics. Rather than focusing solely on technical performance, current research highlights how these models improve decision-making, operational resilience, and sustainability outcomes. Increasingly, attention is shifting from predictive accuracy alone toward explainable and trustworthy AI systems that support human decision-makers and strengthen confidence in automated recommendations (Öz and Üstüner, 2026).

5.1 Water quality forecasting

Machine learning (ML) and deep learning models now underpin real-time forecasting of key water quality variables, dissolved oxygen (DO), temperature, pH, salinity, ammonia, and nitrite. Random Forest (RF) algorithms have proven robust even with sparse or noisy sensor data, achieving near-perfect accuracy (R² ≈ 0.999) in DO prediction and supporting over 6,000 automated corrective interventions, maintaining survival rates above 90% (Baena-Navarro et al., 2025; Zambrano et al., 2021; Swetha et al., 2023). Hybrid deep learning architectures, such as CNN-LSTM, extend predictive precision to multi-parameter forecasting, particularly for nitrogen compounds, though pH prediction remains a challenge due to nonlinear buffering effects (Jongjaraunsuk et al., 2024). The use of AI for predictive analytics allows real-time monitoring and forecasting of environmental conditions, disease outbreaks, and production outcomes. This foresight enables aquaculture practitioners to proactively manage risks and improve operational efficiency, ultimately leading to more sustainable practices. (Sevin and Dikel 2025).

5.2 Feeding optimization

AI-based control frameworks combine ML, reinforcement learning (RL), and mechanistic models to optimize feed management, one of aquaculture’s largest cost and sustainability levers. Vision-based systems detect appetite and adjust rations in real-time through fish detection and counting algorithms (Lee et al., 2013; Atoum et al., 2015). RL-based controllers, including Q-learning, develop adaptive feeding policies that minimize feed waste and ammonia accumulation (Aljehani et al., 2023a). Mechanistic growth models integrate with these intelligent controllers to dynamically balance feed composition and fish weight trajectories, improving growth rates by 13–47% (Li et al., 2022). Such systems demonstrate how predictive control reduces both resource inefficiencies and environmental burdens.

5.3 Growth prediction, harvest scheduling, and mortality modeling

Predictive control frameworks extend from environmental monitoring to biological and economic optimization. Model predictive control (MPC) systems integrate growth trajectories, density effects, and mortality risks to optimize harvest timing and stock management. Compared to traditional PID and bang-bang controllers, MPC reduces mortality by up to 27% and enhances yield predictability (Aljehani et al., 2023b). Decision-support models incorporating partial-harvest strategies alleviate density constraints, increasing both individual growth and system profitability (Yu and Leung, 2006; Yu et al., 2010). These models bridge operational decisions with economic outcomes, reinforcing predictive analytics as a tool for sustainability and profitability.

5.4 Uncertainty quantification and risk-aware planning

Modern aquaculture models increasingly integrate uncertainty analysis to strengthen decision robustness. Monte Carlo simulations estimate pathogen outbreak probabilities (Fu et al., 2015), while nonlinear MPC frameworks apply moving horizon estimation to maintain system stability under incomplete data (Kamali et al., 2023). Economic MPC approaches have achieved up to 41% improvement in cost efficiency by balancing biomass growth with utility expenditure (Patrón and Ricardez-Sandoval, 2024). Integrative modeling reviews emphasize hybrid approaches that merge mechanistic and AI-driven prediction for dynamic, adaptive control of aquaculture systems.

5.5 Explainable forecasting and human-centered AI

As predictive models become increasingly embedded within aquaculture decision-support systems, explainability has emerged as a critical requirement for adoption and governance. While deep learning architectures often outperform traditional statistical approaches, their black-box nature can limit user trust, regulatory acceptance, and operational transparency. Consequently, explainable AI (XAI) techniques are gaining attention as mechanisms for translating model outputs into actionable insights that can be understood and validated by farmers, managers, and policymakers.

Methods such as SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) provide tools for identifying the relative influence of environmental, biological, and operational variables on model predictions. In water-quality forecasting, these approaches can reveal how dissolved oxygen, temperature, feeding intensity, or stocking density contribute to predicted outcomes, thereby enhancing confidence in automated recommendations. Similarly, explainable forecasting frameworks can clarify the drivers of growth predictions, mortality risk estimates, and feeding-control decisions, supporting more transparent and accountable management practices.

From a Human-Centered Artificial Intelligence (HCAI) perspective, explainability is not merely a technical feature but a prerequisite for meaningful human oversight. By enabling stakeholders to understand why a model generated a particular prediction or control action, XAI approaches facilitate collaborative decision-making between AI systems and human operators. Such transparency is particularly important in aquaculture, where decisions affect animal welfare, environmental sustainability, and economic livelihoods. Consequently, future predictive-control systems should be evaluated not only on accuracy and efficiency but also on their ability to provide interpretable, trustworthy, and user-centered decision support.

6 Fish health, biosecurity, and antimicrobial resistance

Effective fish health management underpins sustainable aquaculture by balancing productivity, animal welfare, environmental protection, and food safety. The convergence of artificial intelligence (AI), molecular diagnostics, and biosecurity frameworks is redefining disease management, shifting practice from reactive treatment toward predictive prevention and continuous surveillance. Within a One Health context, this transformation recognizes the interconnectedness of aquatic animal health, human health, and environmental integrity while aligning with the governance objectives of the Food and Agriculture Organization’s (FAO) Blue Transformation agenda. Consequently, fish health management is increasingly viewed not only as a technical challenge but also as a governance issue requiring coordinated action across biological, environmental, institutional, and social systems (Öz et al., 2025).

6.1 Disease detection: imaging, molecular, and AI diagnostics

AI-enabled imaging and molecular diagnostics are transforming early disease detection through improvements in speed, precision, and automation. Multimodal imaging technologies, including RGB imaging, fluorescence imaging, and ultrasonic sensing, combined with convolutional neural networks (CNNs), enable real-time recognition of pathogens such as Ichthyophthirius multifiliis and Flavobacterium columnare, achieving precision rates above 95% (Thakur et al., 2023; Li et al., 2022). These systems complement molecular approaches such as polymerase chain reaction (PCR), loop-mediated isothermal amplification (LAMP), and CRISPR-based diagnostics, which remain gold standards for pathogen confirmation and validation (Jaies et al., 2024; Kumar et al., 2014).

Compared with conventional laboratory diagnostics, AI-based workflows substantially reduce detection times from days to minutes, lower operational costs through automation, and enable continuous in situ monitoring. Such capabilities are particularly important in intensive production systems where rapid disease spread can generate significant economic losses. Nevertheless, these systems depend on high-quality labeled datasets, robust sensor infrastructures, and standardized validation procedures that remain unevenly distributed across aquaculture-producing regions. Developing economies, including the Philippines, continue to face limitations in diagnostic infrastructure and technical capacity, highlighting the need for inter-laboratory benchmarking and harmonized validation frameworks to ensure reliability across species and production systems.

6.2 Outbreak forecasting and spatial epidemiology

Beyond disease detection, AI increasingly supports proactive disease prevention through outbreak forecasting and spatial epidemiology. Predictive models integrate environmental, hydrological, climatic, and host-level data to estimate disease risks and identify emerging hotspots. Geospatial analytics and machine learning approaches have been applied to forecast outbreaks of white spot syndrome virus (WSSV) and tilapia lake virus (TiLV) in Southeast Asia by combining weather data, water-quality information, and farm management records (Redman-White et al., 2023).

These predictive systems support early warning mechanisms and adaptive biosecurity interventions, enabling farmers and regulators to respond before outbreaks escalate. However, forecasting performance depends heavily on data availability, interoperability, and institutional cooperation. Effective disease surveillance therefore requires governance structures that facilitate secure data sharing among farms, laboratories, government agencies, and research institutions. National surveillance initiatives in the Philippines and Vietnam illustrate the potential of public–private partnerships for developing interoperable biosurveillance networks, although fragmented institutional responsibilities and inconsistent data standards remain persistent challenges.

6.3 AI in antimicrobial resistance surveillance

Antimicrobial resistance (AMR) represents one of the most significant governance and public health challenges facing global aquaculture. The sector contributes to environmental reservoirs of antimicrobial-resistant microorganisms that may be transferred among aquatic organisms, terrestrial livestock, wildlife, and humans (Milijasevic et al., 2024). Addressing this challenge requires surveillance systems capable of integrating biological, environmental, and epidemiological data across sectors.

AI-driven tools increasingly support genomic surveillance and predictive modeling of antimicrobial resistance genes, enabling more rational antibiotic use and facilitating the discovery of alternative therapeutics, including antimicrobial peptides and bacteriophage-based interventions (Arnold et al., 2025; Lv et al., 2020). Within a One Health framework, these technologies strengthen national AMR action plans by linking aquaculture surveillance with broader human and animal health monitoring systems. In Southeast Asia, AI-supported advisory platforms have contributed to reductions in antibiotic misuse, including reports of up to a 30% reduction in prophylactic antibiotic use among Philippine shrimp and tilapia farms adopting digital decision-support systems (Bondad‐Reantaso et al., 2023).

Importantly, AMR surveillance demonstrates how AI can function as an integrative governance tool rather than merely a diagnostic technology. By connecting genomic, environmental, production, and health datasets, AI facilitates coordinated responses to emerging resistance threats across multiple sectors and jurisdictions.

6.4 Integration with biosecurity, governance, and data ethics

Sustainable disease management and AMR mitigation require governance systems that extend beyond technological innovation. AI-enabled biosurveillance must operate within transparent governance frameworks that promote accountability, privacy protection, ethical data management, and equitable access to information. The increasing use of farm-level surveillance technologies raises important questions regarding data ownership, consent, cybersecurity, and the responsible sharing of health information, particularly in transboundary trade and disease-reporting contexts.

Effective governance therefore requires harmonized biosecurity policies, farmer training programs, interoperable pathogen databases, and regional coordination mechanisms (Wright et al., 2023; Davies and Wales, 2019). Southeast Asian initiatives, including the ASEAN Guidelines on AMR in Aquaculture, increasingly emphasize multi-stakeholder collaboration among farmers, industry groups, regulators, veterinarians, and researchers. However, implementation effectiveness remains closely linked to institutional capacity, technical infrastructure, and the adoption of interoperable data standards.

From a One Health perspective, governance functions as the connecting mechanism that links technological innovation with public health outcomes. AI systems generate value not only through improved diagnostics but also by strengthening transparency, accountability, and resilience throughout aquatic food systems. Consequently, the long-term success of AI-enabled biosecurity will depend on governance frameworks capable of integrating technological, environmental, and social dimensions of disease management.

6.5 One Health governance perspective

The growing convergence of AI, biosecurity, and antimicrobial resistance surveillance underscores the importance of a One Health governance approach. In this framework, aquatic animal health, environmental health, and human health are treated as interconnected domains supported by shared data infrastructures, coordinated surveillance systems, and collaborative governance mechanisms. AI functions as an enabling layer that integrates molecular diagnostics, biosurveillance, outbreak forecasting, and AMR monitoring into a unified decision-support ecosystem.

Such an approach aligns closely with the FAO Blue Transformation agenda by promoting sustainable production, reducing disease-related losses, minimizing antimicrobial misuse, and strengthening food-system resilience. It also reflects Human-Centered Artificial Intelligence (HCAI) principles by ensuring that AI-supported decisions remain transparent, interpretable, and responsive to the needs of farmers, regulators, and local communities. Ultimately, responsible AI deployment in fish health requires governance systems that balance technological innovation with ethical oversight, stakeholder participation, and environmental stewardship (Figure 5).

Figure 5

7 Genomics, eDNA, and omics–AI in aquaculture

The convergence of genomics, environmental DNA (eDNA), and artificial intelligence (AI) is transforming aquaculture from descriptive biology to predictive bioinformatics. These technologies together enable a shift from population-level management toward individualized, data-informed breeding, health, and sustainability strategies. In the context of the FAO Blue Transformation agenda and One Health frameworks, omics–AI integration provides the analytical backbone for resilient, transparent, and ethically governed aquaculture systems. Beyond technological innovation, however, the increasing value of genomic information raises important questions regarding ownership, access, governance, and the equitable distribution of benefits derived from aquatic genetic resources.

7.1 AI in selective breeding and genotype–phenotype prediction

Aquaculture remains the fastest-growing food production sector globally, where genetic improvement represents a cornerstone for sustainable intensification (Houston et al., 2020). Genomic selection, enhanced by AI, now enables the prediction of economically important traits such as growth, feed efficiency, and disease resistance with unprecedented accuracy. By integrating genomic, transcriptomic, proteomic, and metabolomic datasets, AI algorithms such as deep neural networks and Bayesian models learn to predict phenotypic outcomes directly from multi-omics inputs, accelerating selective breeding cycles (Andersen et al., 2025; Harfouche et al., 2019).

This AI–genomics synergy extends beyond selection. Machine learning models interpret complex genotype-environment interactions and forecast trait performance under climate variability. In advanced systems, genomic selection tools interface with biotechnologies, including CRISPR-Cas gene editing and surrogate broodstock technologies, to design breeding strategies that optimize both productivity and welfare (Houston et al., 2020; Nguyen et al., 2022).

However, the application of such integrated pipelines in Asia, including the Philippines, remains constrained by bioinformatics capacity gaps. Limited local genome repositories, underdeveloped high-performance computing infrastructure, and data fragmentation impede large-scale multi-omics integration. Establishing national genomic databases and AI-ready reference genomes for key species such as Penaeus vannamei and Macrobrachium rosenbergii would significantly strengthen regional breeding programs and genetic conservation efforts.

7.2 Microbiome analysis, productivity, and welfare correlations

The integration of host–microbiome omics data with AI analytics offers new insights into productivity and welfare optimization. High-throughput sequencing and metagenomics now allow the profiling of beneficial microbes, pathogen consortia, and antimicrobial resistance genes in aquatic environments (Canellas et al., 2022; Sundaray et al., 2022). Machine learning models trained on these datasets can detect shifts in microbiome composition associated with stress or disease before visible symptoms emerge, providing an early warning system for health management.

Multi-omics frameworks also advance understanding of the physiological and immunological pathways that underpin resilience and performance. Currently, more than fifty aquaculture species possess draft genomes, enabling the discovery of SNP markers for adaptive and welfare-related traits (Sundaray et al., 2022). The hologenomic approach, which combines host and microbiome genomics, further enhances predictions of growth and immune function (Limborg et al., 2018). When processed through AI-driven bioinformatics pipelines, such data refine selection criteria for disease resistance, feed conversion efficiency, and environmental tolerance, all of which are central to sustainable aquaculture intensification.

7.3 eDNA detection and machine learning trade-offs

Environmental DNA (eDNA) technologies provide a non-invasive and scalable platform for monitoring biodiversity, pathogens, and invasive species in aquaculture systems (Bohara et al., 2022). When coupled with AI, eDNA becomes a predictive tool rather than a passive one. Machine learning algorithms, particularly convolutional and recurrent neural networks, can classify environmental sequences, detect contamination signatures, and forecast ecological risk in near real time (Cordier et al., 2017; Macaulay et al., 2022).

This approach circumvents many limitations of traditional taxonomic identification by linking molecular signals directly to ecological patterns. Low-cost sequencing platforms such as Ion Torrent and MinION further democratize eDNA-based monitoring, enabling farm-level biosurveillance for pathogens including Lepeophtheirus salmonis and Paramoeba perurans (Peters et al., 2018). Nevertheless, challenges persist in data standardization, sequence quality control, and nucleic acid degradation under tropical environmental conditions. These limitations highlight the need for regionally calibrated eDNA–AI models capable of accounting for ecological and climatic variability.

7.4 Ethical, legal, and governance dimensions of genomic data

As genomics and AI become increasingly embedded within aquaculture governance systems, questions of data ownership, access, benefit-sharing, and sovereignty assume growing importance. Genomic and eDNA datasets are frequently derived from local aquatic biodiversity and therefore constitute valuable biological and informational resources. Their collection, storage, and use raise complex ethical and legal questions regarding who owns genetic information, who controls access to it, and who benefits from resulting innovations.

A central governance concern is genomic sovereignty, which refers to the right of nations and communities to exercise authority over genetic resources originating within their territories. For biodiversity-rich countries, including the Philippines and many Southeast Asian nations, genomic datasets derived from endemic species represent strategic assets with scientific, economic, and conservation value. The increasing digitization of genetic resources through AI-ready databases and cloud-based bioinformatics platforms creates new risks of unequal access and potential appropriation by actors possessing greater technological and financial resources.

These concerns intersect with broader debates regarding digital sequence information (DSI) and the implementation of the Nagoya Protocol on Access and Benefit-Sharing. While the protocol was originally designed to regulate access to physical genetic resources, advances in sequencing technologies have created uncertainty regarding ownership and governance of digitized genomic information. Aquaculture increasingly relies on open-access genomic repositories, raising questions about how benefits derived from AI-assisted breeding programs, biomarker discovery, or commercial applications should be shared with source countries and communities.

The governance challenge extends beyond national sovereignty to include the rights of Indigenous Peoples and local communities whose traditional ecological knowledge and stewardship practices contribute to biodiversity conservation. Indigenous and community-managed aquatic ecosystems often contain unique genetic resources that may become targets for genomic research and commercial development. Responsible AI-enabled genomics therefore requires recognition of Indigenous biodiversity rights, meaningful stakeholder participation, prior informed consent, and equitable benefit-sharing mechanisms that acknowledge local contributions to resource stewardship.

From a Human-Centered Artificial Intelligence (HCAI) perspective, genomic governance must prioritize transparency, accountability, and inclusivity. Open-access genomic databases offer substantial scientific benefits but must be balanced against safeguards that prevent exploitation, ensure fair attribution, and promote equitable participation in innovation systems. Governance frameworks should therefore integrate biodiversity conservation objectives, data sovereignty principles, and ethical AI standards to ensure that advances in genomics support both technological progress and social justice.

Investing in bioinformatics education, data ethics training, and cross-sector collaboration will be critical for enabling countries to participate meaningfully in the genomics–AI ecosystem. Equally important is the development of regional governance mechanisms capable of balancing scientific openness with biodiversity protection and benefit-sharing obligations.

7.5 Critical analysis

AI-driven genomics, microbiomics, and eDNA analytics are redefining aquaculture from descriptive monitoring toward predictive and preventive management. However, the principal barriers to adoption are increasingly institutional rather than technical. While advances in sequencing technologies and AI algorithms continue to improve analytical capabilities, governance frameworks have not evolved at the same pace.

Developed economies possess significant advantages in genomic infrastructure, computational capacity, and access to large reference databases, enabling rapid progress in AI-assisted breeding and bioinformatics. In contrast, many Asian aquaculture-producing nations continue to face limitations in genomic infrastructure, data governance capacity, and bioinformatics expertise. Without deliberate policy interventions, these disparities risk creating new forms of digital and genomic inequality.

The future of omics–AI integration will therefore depend on more than technological innovation alone. Sustainable implementation requires robust genomic infrastructures, interoperable data ecosystems, transparent governance arrangements, and equitable benefit-sharing mechanisms. Particular attention must be given to genomic sovereignty, Indigenous biodiversity rights, and Nagoya Protocol obligations to ensure that the benefits of genomic innovation are shared fairly among countries, communities, and stakeholders.

Ultimately, the fusion of omics intelligence and AI represents both a scientific frontier and a governance frontier. Its long-term success will depend on the ability of aquaculture systems to balance innovation with stewardship, openness with sovereignty, and scientific advancement with social and ecological responsibility.

8 Remote sensing, GIS, and AI-driven decision support in aquaculture

Remote sensing (RS) and Geographic Information Systems (GIS) have become indispensable tools for spatial intelligence and adaptive governance in aquaculture, providing a framework that integrates environmental, ecological, and socio-economic data into decision-support systems. The convergence of satellite and drone analytics with artificial intelligence (AI) enables high-resolution, multi-temporal monitoring of aquaculture environments, transforming site selection, risk assessment, and climate resilience planning from reactive processes into predictive and adaptive management systems (Padmanaban and Sudalaimuthu, 2012; Shaginimol et al., 2023).

At the core of this integration lies the capacity of RS–GIS systems to synthesize diverse datasets, including sea surface temperature, chlorophyll-a concentration, turbidity, bathymetry, precipitation, and coastal land-use patterns, to delineate suitable aquaculture zones. Modern satellite constellations such as Landsat 8, Sentinel-2, and MODIS generate spectral indicators that serve as proxies for water quality, ecosystem productivity, and coastal geomorphology, while unmanned aerial systems provide finer-scale data for habitat mapping and infrastructure monitoring (Snyder et al., 2017). Through multi-criteria spatial analysis, these tools help determine optimal culture sites based on environmental suitability, accessibility, carrying capacity, and socio-economic viability, thereby supporting science-based aquaculture expansion and sustainable coastal planning (Subramani et al., 2017; Ayisi and Apraku, 2016).

Beyond operational efficiency, AI-enabled spatial intelligence increasingly contributes to climate adaptation and resilience planning. By integrating long-term environmental observations with predictive analytics, RS–GIS systems support anticipatory governance approaches that help aquaculture producers adapt to changing climatic conditions, extreme weather events, and ecosystem shifts. This capability aligns closely with the Food and Agriculture Organization’s (FAO) Blue Transformation agenda and the objectives of Sustainable Development Goal (SDG) 14 (Life Below Water), which emphasize sustainable aquatic food production, ecosystem protection, and climate-resilient management of marine and coastal resources.

8.1 AI-enhanced monitoring of harmful algal blooms and climate hazards

A major advancement in recent years is the integration of AI with RS–GIS systems for predictive modeling of harmful algal blooms (HABs) and climate-related risks. Machine learning algorithms trained on ocean-color and meteorological data can forecast bloom formation, identify high-risk phytoplankton communities (e.g., Pseudo-nitzschia and Karenia), and support early warning systems with near-real-time accuracy (Smith and Bernard, 2020; Davidson et al., 2021). Spectral indices derived from chlorophyll-a concentrations and sea surface temperatures are increasingly processed through neural networks to estimate bloom probability and spatial extent, providing aquaculture operators with actionable alerts that enhance farm-level preparedness.

Beyond HAB monitoring, AI-integrated RS data support risk forecasting for typhoons, flooding, storm surges, sea-level rise, coastal erosion, and extreme precipitation events, all of which are expected to intensify under climate change. Combining satellite-derived precipitation, sea-level, and topographic information with hydrodynamic models allows early identification of vulnerable aquaculture sites and supports climate-resilient zoning strategies. Such applications move aquaculture governance beyond disaster response toward anticipatory adaptation planning and strengthen the capacity of coastal communities to withstand environmental shocks.

8.2 Case applications and regional initiatives

In Southeast Asia, a growing number of initiatives illustrate the operationalization of RS–GIS–AI integration. The Southeast Asian Fisheries Development Center (SEAFDEC) employs satellite-derived sea surface temperature and chlorophyll data to monitor mariculture zones and predict HAB dynamics in the Gulf of Thailand and the Philippines. The National Mapping and Resource Information Authority (NAMRIA) provides foundational geospatial datasets, including bathymetric, hydrographic, and land-cover maps, that support aquaculture suitability modeling conducted by the Bureau of Fisheries and Aquatic Resources (BFAR). BFAR’s PhilFARM system integrates GIS layers with farm registration and environmental monitoring databases to facilitate spatially explicit management of aquaculture zones.

Collectively, these initiatives demonstrate how geospatial technologies can strengthen adaptive governance by linking environmental monitoring, resource planning, and regulatory oversight. They also highlight the growing role of public–private partnerships and interagency collaboration in building climate-responsive aquaculture management systems throughout the region.

8.3 Modeling carrying capacity and water dynamics

Coupling RS-derived datasets with hydrodynamic models and machine learning further strengthens environmental planning by estimating water-exchange rates, nutrient loading, assimilative capacity, and ecological carrying capacity. In Japan, GIS-based models identified 74% of southern Hokkaido’s shallow marine zones as optimal for kelp cultivation, emphasizing suspended solids as a critical parameter influencing productivity (Radiarta et al., 2011). Similarly, GIS multi-criteria analyses in India and Africa have demonstrated the value of integrating biophysical and socio-economic variables to optimize aquaculture site selection and resource allocation (Subramani et al., 2017; Quansah et al., 2007).

These examples underscore the scalability of RS–GIS frameworks when reinforced by AI-assisted environmental prediction. More importantly, carrying-capacity models provide a foundation for balancing production objectives with ecosystem sustainability, helping prevent over-intensification and environmental degradation in coastal and inland aquaculture systems.

8.4 Climate-adaptive spatial governance and sustainability planning

The growing availability of geospatial intelligence is transforming aquaculture governance from static zoning approaches toward dynamic, climate-adaptive planning. AI-assisted RS–GIS systems enable continuous monitoring of environmental change, allowing policymakers to adjust site allocations, production limits, and risk-management strategies in response to evolving ecological conditions.

This transition is particularly significant in the context of climate resilience. By identifying climate-sensitive production areas, forecasting environmental risks, and supporting adaptive management interventions, spatial intelligence systems contribute directly to resilience-building efforts among coastal communities. These capabilities support the objectives of the FAO Blue Transformation agenda by promoting sustainable intensification while reducing environmental vulnerability. They also contribute to SDG 14 by improving ecosystem stewardship, protecting aquatic biodiversity, and strengthening the sustainability of aquatic food systems.

From a governance perspective, climate-adaptive spatial intelligence supports evidence-based policymaking, participatory planning, and integrated coastal management. When combined with local ecological knowledge and stakeholder engagement processes, AI-enabled spatial decision-support systems can facilitate more inclusive and socially responsive approaches to aquaculture development.

8.5 Limitations and the path toward inclusive spatial governance

Despite their transformative potential, RS–GIS–AI applications face persistent constraints. Cloud cover and atmospheric interference limit optical satellite observations in tropical regions; high-resolution imagery remains costly for many smallholder producers; and data latency can reduce the timeliness of decision-making. Furthermore, integrating heterogeneous datasets requires robust computational infrastructure, geospatial expertise, and institutional capacity, all of which remain unevenly distributed across Southeast Asian aquaculture economies.

Addressing these challenges requires sustained investment in open-access geospatial platforms, digital infrastructure, technical training, and public–private partnerships capable of democratizing access to spatial intelligence. Equally important is the development of governance frameworks that ensure transparency, interoperability, and equitable participation in geospatial decision-making processes.

Ultimately, the fusion of remote sensing, GIS, and AI is transforming aquaculture into a knowledge-based industry grounded in spatial precision, environmental stewardship, and climate resilience. By coupling spectral observations with predictive analytics and participatory governance, aquaculture systems can evolve toward climate-smart, ethically governed, and economically inclusive models that support the FAO Blue Transformation agenda, contribute to SDG 14, and strengthen the resilience of aquatic food systems under a changing climate.

The overall interaction among these technologies is visually summarized in Figure 6. The figure illustrates the interconnected flow of data from satellite sensors, drones, and in situ monitoring systems toward a central AI analytics hub, which processes multi-source datasets to generate suitability maps, early warning systems, climate-risk forecasts, and decision-support dashboards. This architecture represents a closed-loop governance model in which real-time environmental observations continuously refine predictive models, enabling adaptive, climate-resilient, and evidence-based aquaculture management.

Figure 6

9 Supply chain, safety, and traceability in aquaculture

As aquaculture supply chains become increasingly globalized, ensuring food safety, traceability, and transparency has emerged as a strategic priority for producers, regulators, retailers, and consumers. Modern aquaculture products often pass through complex networks of farms, processors, distributors, exporters, and retailers before reaching end users. These extended supply chains create challenges related to product authentication, contamination control, regulatory compliance, and sustainability verification. Artificial intelligence (AI), computer vision (CV), blockchain, Internet of Things (IoT) technologies, and advanced analytics are increasingly deployed to address these challenges, transforming supply-chain management from a reactive process into a predictive and governance-oriented system.

Beyond operational efficiency, digital traceability systems contribute to broader sustainability objectives by improving accountability, reducing fraud, strengthening food safety oversight, and supporting responsible aquatic food production. In this context, AI-enabled traceability serves not only as a technological innovation but also as a governance mechanism that aligns production practices with consumer expectations, regulatory requirements, and sustainability goals.

9.1 Computer vision and AI for safety monitoring and defect detection

Computer vision and deep learning technologies have become important tools for monitoring infrastructure integrity, product quality, and food safety across aquaculture supply chains. Automated inspection systems can identify structural defects in aquaculture nets, including biofouling, vegetation growth, and physical damage, improving detection precision by 3.69–6.58% across multiple datasets (Akram et al., 2023). Deep-learning architectures such as Faster R-CNN, SSD, and YOLOv3 have been successfully applied to biofouling detection, helping maintain water circulation, fish welfare, and production efficiency (Paraskevas and Kavallieratou, 2022).

AI applications extend beyond infrastructure monitoring. Bayesian network models have demonstrated the ability to forecast food-safety vulnerabilities with approximately 81% accuracy using expert-informed risk assessments (Marvin et al., 2020). Similarly, deep learning approaches support species identification, product classification, stock enumeration, and water-quality monitoring, enabling continuous surveillance throughout production and distribution systems (Sun et al., 2020).

Food-safety monitoring remains particularly important because aquaculture products may be exposed to contaminants including methylmercury, polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers (PBDEs), Salmonella, mycotoxins, drug residues, heavy metals, and agricultural chemicals originating from feed ingredients and environmental sources (Mantovani et al., 2015; Tacon and Metian, 2008). Although vegetable-based feed formulations can reduce contaminant accumulation, they may also influence nutritional quality and product composition (Mantovani et al., 2015). Consequently, Hazard Analysis Critical Control Point (HACCP) systems, routine surveillance programs, and origin verification procedures remain essential components of contemporary food-safety governance (Lie, 2008).

9.2 Blockchain–AI integration for traceability and supply-chain governance

Traceability has emerged as one of the most prominent applications of digital technologies in aquaculture. Mislabeling affects nearly 20% of global seafood products, creating economic losses, undermining consumer confidence, and complicating regulatory oversight (Hisham et al., 2025). Blockchain technology has been proposed as a mechanism for strengthening trust and transparency by creating immutable records of production, processing, transportation, and retail transactions.

However, blockchain alone does not generate value from traceability data. Its greatest potential emerges when integrated with AI, IoT sensors, and digital identity systems. In a blockchain–AI framework, IoT devices collect environmental and production data, blockchain platforms provide secure and auditable record storage, and AI systems analyze accumulated information to identify anomalies, forecast risks, optimize logistics, and support compliance monitoring. This architecture transforms traceability systems from passive repositories into active decision-support platforms capable of generating real-time insights.

The Philippine ‘Tracy’ project provides an example of this evolving paradigm. The platform utilizes smartphone-based data collection to record harvest information, vessel identities, and transaction histories that can be linked to broader traceability systems (Shamsuzzoha et al., 2023). Similar initiatives demonstrate how digital traceability can improve transparency while supporting market access and regulatory compliance. Nevertheless, implementation challenges remain substantial, particularly among smallholder producers facing constraints related to infrastructure, technical expertise, legal frameworks, and financial resources (Dasaklis et al., 2022; Gapparov et al., 2025).

To better understand technology choices, Table 4 compares blockchain systems with conventional database architectures used in supply-chain management.

As shown in Table 4, blockchain offers important advantages in transparency, auditability, and trust among distributed stakeholders. However, conventional databases frequently remain more cost-effective, scalable, and easier to maintain. Consequently, blockchain should not be viewed as a universal solution but rather as one component within a broader digital-governance ecosystem.

9.3 AI for demand forecasting, price stabilization, and logistics optimization

AI increasingly supports decision-making across downstream supply-chain operations, including market forecasting, inventory management, transportation planning, and cold-chain optimization. By integrating machine learning, IoT technologies, and predictive analytics, aquaculture enterprises can improve fish health management, optimize feeding strategies, and enhance production planning while reducing dependence on manual monitoring (Rather et al., 2024).

Market intelligence applications are particularly important for reducing economic uncertainty. AI-supported forecasting systems help producers anticipate fluctuations in demand and adjust production schedules accordingly, contributing to price stabilization for aquaculture products relative to wild-capture fisheries (Dahl, 2017). Intelligent traceability systems that integrate wireless sensor networks with statistical process control methods further support continuous monitoring of storage and transportation conditions. Factors such as temperature exposure, precooling procedures, transport duration, and logistics routes have direct implications for product quality, shelf life, and consumer safety (Mai, 2010; Xiao et al., 2016).

Economic analyses highlight the significance of these improvements. Procurement activities account for more than 70% of total aquaculture supply-chain costs, while transportation expenses represent approximately 43.57% of costs within wild-capture fisheries supply chains. AI-assisted logistics optimization can therefore generate substantial savings through improved route planning, inventory management, transport selection, and cold-chain efficiency (Guritno and Tanuputri, 2017).

9.4 Cost–benefit, scalability, and governance considerations

Despite the growing interest in blockchain and AI-enabled traceability, implementation decisions should be guided by cost–benefit considerations rather than technological enthusiasm. For many small-scale producers, the costs associated with blockchain infrastructure, digital identity systems, data integration, and personnel training may exceed short-term economic benefits. Consequently, the most effective deployments often occur in high-value export-oriented supply chains where traceability delivers measurable market advantages.

Scalability presents another significant challenge. As transaction volumes increase, blockchain systems may experience constraints related to processing speed, storage requirements, interoperability, and energy consumption. Hybrid architectures that combine conventional databases for operational data management with blockchain layers for critical traceability records may therefore offer a more practical pathway for large-scale implementation.

Governance considerations are equally important. Traceability systems generate commercially sensitive information regarding production practices, trade relationships, pricing structures, and environmental performance. Questions concerning data ownership, privacy, access rights, cybersecurity, and accountability therefore become central to digital supply-chain governance. Ethical AI principles should guide system design to ensure transparency, explainability, fairness, and accessibility, particularly for smallholders and cooperatives that may otherwise be excluded from digital innovation ecosystems.

9.5 Strategic roadmap for digital traceability

Successful implementation of AI-enabled traceability systems requires a phased and governance-oriented approach:

• Pilot implementation using high-value species and limited supply chains to validate technical and economic feasibility.

• Integration with existing infrastructure, including IoT devices, mobile applications, and enterprise databases.

• Capacity building for producers, cooperatives, and regulators to improve digital literacy and data management skills.

• Regulatory harmonization to ensure compliance with food-safety, trade, labeling, and environmental standards.

• Scalability planning based on demonstrated value creation and stakeholder acceptance.

• Governance development addressing data ownership, privacy protection, cybersecurity, and accountability.

Table 5 summarizes the principal digital technologies currently applied across aquaculture supply chains and their associated sustainability contributions.

As summarized in Table 5, the greatest benefits emerge not from individual technologies operating independently but from integrated digital ecosystems that combine monitoring, analytics, traceability, and governance functions.

10 Human-centered artificial intelligence and user adoption in digital aquaculture

Human-centered artificial intelligence (HCAI) has emerged as a guiding paradigm for the integration of advanced computational systems into aquaculture and smart farming contexts, emphasizing augmentation of human intelligence rather than its displacement (Holzinger et al., 2022, 2024). Central to this paradigm is the notion that effective AI deployment requires the symbiotic interaction between human intuition and machine precision, particularly in data-driven aquaculture systems where decision complexity, environmental variability, and ethical accountability intersect. Explainable artificial intelligence (XAI) plays a critical role in establishing user confidence by offering transparency, interpretability, and actionable insights, thus ensuring that algorithmic recommendations align with practitioners’ expertise and contextual realities (Liao and Varshney, 2021).

However, the integration of AI into user-facing systems presents dual challenges. On one hand, automation enables enhanced efficiency, precision, and scale; on the other, excessive reliance on automation risks diminishing human empathy, situational awareness, and professional agency (Lu et al., 2024). The design of HCAI frameworks therefore prioritizes reliability, safety, and human mastery, encouraging systems that strengthen self-efficacy and creative problem-solving (Shneiderman, 2020; Barmer et al., 2021). Effective implementation depends on adaptive, context-aware mechanisms that enable human oversight and bias mitigation (Barmer et al., 2021).

Despite significant technological progress, widespread adoption of HCAI and digital aquaculture tools remains constrained by persistent gaps in digital literacy and institutional readiness. These skill deficits are particularly evident among older populations and regions with limited access to digital training infrastructure (Acharya, 2025; Bogoslov et al., 2024). Traditional academic curricula often underrepresent interdisciplinary skills such as AI literacy, human-computer interaction (HCI), and qualitative data analysis—competencies vital to managing the hybrid intelligence systems underpinning modern aquaculture (Lu et al., 2024; Zhang et al., 2024). Bridging this gap requires institutional reforms that strengthen faculty development, invest in AI infrastructure, and promote academia–industry collaboration for experiential, hands-on learning.

Adoption barriers extend beyond skills to include trust, affordability, connectivity, and regulatory constraints. Users must understand system limitations and maintain informed engagement with AI interfaces to prevent overdependence or misuse (Shneiderman, 2020). Explainable AI provides a mechanism for aligning automation with user comprehension and contextual needs, promoting balanced adoption that preserves accountability and ethical control (Liao and Varshney, 2021).

Regional adoption studies reinforce these dynamics. In the United Arab Emirates, 80% of users reported improved task efficiency with AI chatbots, underscoring the importance of natural language processing in user acceptance (Ismail, 2024). Conversely, comparative analyses between European and Latin American healthcare sectors revealed differing adoption priorities: while European practitioners emphasized quality-of-life metrics, Latin American users focused on treatment adherence and complication management (Puig-Bosch et al., 2024). These findings highlight that cultural context, domain-specific trust, and socioeconomic conditions shape HCAI adoption, underscoring the need for context-sensitive deployment strategies across aquaculture and allied bioindustries.

10.1 Economic feasibility and return on investment

The economic feasibility of AI in aquaculture depends on whether productivity gains, reduced mortality, improved feed conversion, labor savings, and market benefits outweigh capital and operating costs. High-value species, intensive systems, export-oriented farms, and large commercial operations are more likely to achieve positive returns because they can distribute hardware, software, maintenance, and training costs across larger production volumes. In contrast, smallholder farms may experience delayed or limited return on investment due to high upfront costs, limited internet connectivity, lack of technical support, and dependence on low-margin production systems.

Feed optimization and disease prevention represent the strongest economic entry points because feed is one of the largest production costs and disease outbreaks can generate severe financial losses. AI-based feeding systems may improve profitability when they reduce feed waste, improve feed conversion ratio, and stabilize growth performance. Similarly, early-warning systems for disease and water-quality deterioration may generate economic value by preventing mortality events. However, these benefits are not automatic. Poor sensor calibration, model drift, false alarms, incorrect predictions, and insufficient farmer training can reduce or eliminate expected gains.

Therefore, AI adoption should follow a phased investment model: low-cost monitoring and mobile decision-support tools for smallholders; integrated sensor and feeding-control systems for semi-intensive farms; and advanced AI, digital twins, computer vision, and automated control systems for intensive commercial farms. Cost–benefit assessment should include hardware acquisition, installation, calibration, cloud services, software subscriptions, cybersecurity, maintenance, staff training, downtime risk, and replacement costs. This approach avoids technology-driven adoption and instead promotes economically justified, scale-appropriate implementation.

11 Ethics, governance, and policy dimensions of AI integration

As digital aquaculture advances toward large-scale data integration, ethical and governance challenges become central to ensuring equitable, transparent, and sustainable innovation. Algorithmic bias remains one of the most critical risks in AI systems, stemming from biased training datasets, structural inequalities, or flawed algorithmic logic (Isley, 2022; Pendharkar, 2023). Manifestations include proxy discrimination, disparate treatment, and feature-based bias, which distort decision-making processes and undermine perceptions of fairness (Wang et al., 2024; Kordzadeh and Ghasemaghaei, 2021). Addressing these concerns requires interdisciplinary cooperation among technologists, policymakers, and ethicists to establish transparent audit mechanisms, preventive controls, and algorithmic accountability frameworks. Comparative governance models show marked variation: the United States emphasizes preventive regulation and liability, while the European Union, Canada, and Australia employ stewardship-based frameworks (Wang et al., 2024).

In parallel, data ownership, privacy, and biosecurity represent a second axis of ethical concern. The digitization of biological systems, including genetic, behavioral, and health data, poses novel risks encompassing pathogen transmission, bioengineering misuse, and unauthorized data access (DiEuliis et al., 2018). Current regulatory instruments such as the Common Rule and HIPAA provide partial safeguards through deidentification and encryption but remain insufficient for bio-digital convergence contexts (Kerr et al., 2017; Arellano et al., 2018). Governance evolution in biosciences now reflects a shift from purely rights-based models to hybrid frameworks balancing privacy, innovation, and economic interests (Hockings, 2016). This transition demands deliberative, participatory policymaking that accounts for public values and inclusivity.

Digital transformation in aquaculture is further linked to the Sustainable Development Goals (SDGs), particularly those addressing poverty (SDG 1), hunger (SDG 2), and climate action (SDG 13). Smart aquaculture systems and Industry 4.0 technologies contribute to productivity and sustainability but also risk exacerbating inequality, job displacement, and environmental burdens (FAO, 2022). Adaptive governance approaches, ranging from laissez-faire to precautionary and stewardship models, must align digital innovation with social equity and environmental integrity (Linkov et al., 2018). Regional examples, such as Ceará’s smart governance model in Brazil, illustrate how digital tools can be leveraged to enhance resilience among vulnerable populations while advancing sustainability objectives (Furtado et al., 2023).

Public–private data collaboratives exemplify this intersection of innovation and ethics. With more than 150 initiatives globally by 2017, these partnerships mobilize private sector data for public good but face persistent challenges of data privacy, equitable benefit sharing, and transparent governance (Young et al., 2019; Bak et al., 2025). Although over 80 AI ethics documents have been published since 2016, the field remains hindered by homogeneity in authorship and limited policy translation (Schiff et al., 2019). The resulting gap between innovation and regulation necessitates new participatory benchmarks to align ethical intent with technological capability (Davis, 2014). The major themes, implementation barriers, and strategic directions for AI-driven aquaculture toward 2030 are synthesized in Table 6.

12 Emerging case studies and research frontiers (2025–2030)

Recent applications and pilot studies demonstrate the transformative potential, and persistent challenges, of integrating AI into aquaculture. Stereo-vision systems have revolutionized phenotypic monitoring, enabling accurate 3D assessments of biomass, distribution, and behavior across aquatic species (Zhao et al., 2025). In commercial sea cages, stereovision imaging revealed that larger Atlantic salmon occupy deeper layers while smaller individuals aggregate near the surface, confirming that behavioral stratification affects biomass estimation (Sauphar et al., 2023). Persistent limitations include detection of ectoparasites and welfare indicators in turbid, high-density environments (Fitzgerald et al., 2025).

Parallel developments in IoT-enabled multi-sensor farms illustrate how integrated data architectures, combining hyperspectral, LIDAR, thermal, and stereo imaging, can optimize environmental monitoring and production efficiency (Rilling et al., 2017; Taylor et al., 2013). In developing economies, IoT deployment has improved yields while reducing input costs, as evidenced by mobile-controlled irrigation systems in India and precision agriculture initiatives in Africa (Garg et al., 2025). Singapore’s sensor-integrated farms further demonstrate how automation and analytics enhance climate resilience and investment viability (Montesclaros et al., 2019).

Pilot studies in emerging industries reveal the importance of readiness and learning curves in technology diffusion. Early adopters of 3D fashion and digital health technologies streamlined workflows but struggled with regulatory uncertainty and multi-stakeholder coordination (Liu and Cui, 2024; Lim and Anderson, 2016). Startups frequently prioritize rapid iteration and market testing over comprehensive methodology, reflecting an agile yet fragmented innovation landscape (Souza et al., 2019; Laporte and O'Connor, 2016).

Looking ahead, robust benchmarking and model transparency are essential for scaling AI in aquaculture. Datasets such as AQUA20 provide critical infrastructure for evaluating deep learning model performance under real-world aquatic conditions (Fuad et al., 2025). Meanwhile, welfare monitoring remains a key research frontier; attention-based architectures such as YOLOv8-CBAM have demonstrated precision exceeding 95% in complex farm environments (Araújo et al., 2025). Future priorities include edge computing, lightweight architectures, federated learning, and multimodal data fusion to enhance efficiency and scalability (Hossain et al., 2024; Wu et al., 2025; Aung et al., 2024).

Finally, the grand challenge for 2025–2030 lies in multi-objective optimization that balances profitability, animal welfare, and environmental sustainability. With aquaculture now surpassing capture fisheries in many regions and encompassing more than 450 species (Mair et al., 2023; Troell et al., 2023), integrative frameworks must reconcile trade-offs across ecological and socio-economic dimensions. Circular economy innovations, such as waste-to-feed systems and microalgae-based bioremediation, hold promise for reducing nutrient discharge and energy use (Dasari et al., 2025). Policy sandboxes aligned with FAO and SDG guidelines can facilitate adaptive governance for responsible digital aquaculture (Little and Mackenzie, 2023).

13 Practical implementation pathway

A practical pathway for responsible AI-enabled aquaculture can be organized into five steps. First, farms should conduct a digital readiness assessment covering connectivity, power supply, sensor availability, staff capacity, and data-management practices. Second, users should prioritize high-impact problems such as feed waste, disease risk, water-quality instability, mortality, labor bottlenecks, or traceability gaps. Third, AI tools should be introduced through small pilot applications before full-scale deployment. Fourth, systems should be validated under real farm conditions, including seasonal variability, turbidity, biofouling, sensor drift, and species-specific behavior. Fifth, scaling should proceed only when the technology demonstrates technical reliability, economic value, farmer usability, and governance safeguards for data ownership, privacy, transparency, and accountability.

For smallholders, the most practical entry points are low-cost sensors, mobile advisory systems, digital record-keeping, and cooperative data-sharing models. For commercial farms, higher-level applications may include automated feeding, computer vision, digital twins, disease-risk forecasting, and supply-chain traceability. For policymakers, priority actions include open data standards, farmer training, public–private pilot programs, digital infrastructure investment, and governance frameworks that prevent exclusion of small-scale producers.

14 Conclusion

Artificial intelligence is transforming aquaculture from a predominantly reactive production system into a predictive, data-driven, and increasingly autonomous sector. This review demonstrates that AI applications now extend across the entire aquaculture value chain, including environmental monitoring, biomass estimation, disease surveillance, feeding optimization, genomics, traceability, and decision support. Beyond technological advances, the synthesis highlights that sustainable digital transformation depends equally on governance structures, institutional capacity, and stakeholder engagement. Three major insights emerge from this review. First, the future of aquaculture will increasingly rely on integrated AI ecosystems that combine IoT networks, computer vision, predictive analytics, digital twins, and omics-based intelligence. Second, the effectiveness and legitimacy of these systems depend on human-centered design principles that prioritize transparency, explainability, trust, and meaningful human oversight. Third, governance challenges, including data ownership, algorithmic bias, cybersecurity, genomic sovereignty, and equitable access to innovation, must be addressed through adaptive regulatory frameworks and collaborative multi-stakeholder approaches. However, despite promising advances, many AI applications remain at the pilot or early-commercial stage, and their broader implementation continues to be constrained by high infrastructure and maintenance costs, sensor reliability, data quality, interoperability limitations, and the need for skilled technical support, particularly in resource-limited farming systems. The proposed AI–Sustainability–Governance Framework provides a conceptual pathway for aligning technological innovation with environmental stewardship, social inclusion, and economic resilience. Looking toward 2030, future research should prioritize large-scale commercial validation, economic feasibility and return-on-investment analyses, federated learning, explainable AI, open benchmarking platforms, digital twins, and responsible data governance mechanisms capable of supporting both innovation and accountability. Ultimately, the success of AI in aquaculture will not be determined solely by advances in algorithms or computing power, but by society’s ability to integrate technology within ethical, inclusive, economically viable, and sustainability-oriented governance systems that support the broader objectives of Blue Transformation and global food security.

References

• 1

  AbidA.DupontC.GallF.ThirdA.KaneF. (2019). “ Modelling data for a sustainable aquaculture”, in: 2019 Global Internet of Things Summit (GIoTS), IEEE1–6. doi: 10.1109/GIOTS.2019.8766376

  • CrossRef
  • Google Scholar

• 2

  AcharyaS. N. (2025). Bridging the skills gap: Role of AI to transform talent acquisition and education in a digitally transformed world. J. Inf. Syst. Eng. Manage.10, 11665. doi: 10.52783/jisem.v10i55s.11665

  • CrossRef
  • Google Scholar

• 3

  AgostiniV. N.OlsenE.TiffayC.AllisonE. H.CojocaruA.CostelloC.et al. (2025). Charting a science course for the sustainable transformation of aquatic food systems. ICES J. Mar. Sci.82(1), fsae184. doi: 10.1093/icesjms/fsae184

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  AkramW.DinM. U.SoudL. S.HussainI. (2025). A review of generative AI in aquaculture: Foundations, applications, and future directions for smart and sustainable farming. arXiv. doi: 10.48550/arXiv.2507.11974

  • CrossRef
  • Google Scholar

• 5

  AkramW.HassanT.ToubarH.AhmedM.MiškovićN.SeneviratneL.et al. (2023). Aquaculture defects recognition via multi-scale semantic segmentation. Expert Syst. Appl.224, 121197. doi: 10.1016/j.eswa.2023.121197

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  AljehaniF.N’DoyeI.Laleg‐KiratiT. (2023a). Feeding control and water quality monitoring in aquaculture systems: Opportunities and challenges. arXiv. doi: 10.48550/arXiv.2306.09920

  • CrossRef
  • Google Scholar

• 7

  AljehaniF.N’DoyeI.Laleg‐KiratiT. (2023b). “ Fish growth tracking and mortality monitoring: Control design and comparisons”, in: Proceedings of the American Control Conference (ACC). Piscataway, NJ: IEEE10156119. doi: 10.23919/ACC55779.2023.10156119

  • CrossRef
  • Google Scholar

• 8

  AndersenL. K.ThompsonN. F.AbernathyJ. W.AhmedR. O.AliA.Al-TobaseiR.et al. (2025). Advancing genetic improvement in the omics era: Status and priorities for United States aquaculture. BMC Genomics26, 11247. doi: 10.1186/s12864-025-11247-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  AraújoV. M.RiliI.GisigerT.GambsS.VasseurE.CellierM.et al. (2025). AI-powered cow detection in complex farm environments. Smart Agric. Technol. 100770. doi: 10.1016/j.atech.2025.100770

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  ArellanoA. M.DaiW.WangS.JiangX.Ohno-MaChadoL. (2018). Privacy policy and technology in biomedical data science. Annu. Rev. Biomed. Data Sci.1, 45–70. doi: 10.1146/annurev-biodatasci-080917-013416

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  ArnoldA.McLellanS.StokesJ. M. (2025). How AI can help us beat AMR. NPJ Antimicrobials Resistance1, 85. doi: 10.1038/s44259-025-00085-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  AtoumY.SrivastavaS.LiuX. (2015). Automatic feeding control for dense aquaculture fish tanks. IEEE Signal Process Lett.22, 1089–1093. doi: 10.1109/LSP.2014.2385794

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  AungT.Abdul RazakR.Bin Md NorA. R. (2025). Artificial intelligence methods used in various aquaculture applications: A systematic literature review. J. World Aquacult Soc. 56(1), e13107. doi: 10.1111/jwas.13107

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14

  AyisiC. L.AprakuA. (2016). The use of remote sensing and GIS in aquaculture: Recent advances and future opportunities. Int. J. Aquaculture6, 1–8. doi: 10.5376/ija.2016.06.0019

  • CrossRef
  • Google Scholar

• 15

  Baena-NavarroR.Carriazo-ReginoY.Torres-HoyosF.Pinedo-LópezJ. (2025). Intelligent prediction and continuous monitoring of water quality in aquaculture: Integration of machine learning and Internet of Things for sustainable management. Water17, 82. doi: 10.3390/w17010082

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  BakM.HorbachD.BuyxA.McLennanS. (2025). A scoping review of ethical aspects of public-private partnerships in digital health. NPJ Digital Med.8, 115. doi: 10.1038/s41746-025-01515-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  BarmerH.DzombakR.GastonM.PalatV.RednerF.SmithC. J.et al. (2021). Human-centered AI. IEEE Pervasive Comput. doi: 10.1184/R1/16560183.V1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  BogoslovI.CormanS.LunguA. (2024). Perspectives on artificial intelligence adoption for European Union elderly in the context of digital skills development. Sustainability16, 4579. doi: 10.3390/su16114579

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  BoharaK.YadavA.JoshiP. (2022). Detection of fish pathogens in freshwater aquaculture using eDNA methods. Diversity14, 1015. doi: 10.3390/d14121015

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  Bondad‐ReantasoM.MacKinnonB.KarunasagarI.FridmanS.Alday‐SanzV.BrunE.et al. (2023). Review of alternatives to antibiotic use in aquaculture. Rev. Aquacult.15, 1123–1149. doi: 10.1111/raq.12786

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  BoydC.McNevinA. A.DavisR. P. (2022). The contribution of fisheries and aquaculture to the global protein supply. Food Secur.14, 915–931. doi: 10.1007/s12571-021-01246-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  BurkeM.NikolićD.FabryP.RishiH.TelferT.Rey PlanellasS. (2025). Precision farming in aquaculture: Non-invasive monitoring of Atlantic salmon (Salmo salar) behaviour in response to environmental conditions in commercial sea cages for health and welfare assessment. Front. Rob. AI12, 1574161. doi: 10.3389/frobt.2025.1574161

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  CanellasA. L. B.CostaW. F.Freitas-SilvaJ.LopesI. R.de OliveiraB. F. R.LaportM. (2022). In sickness and in health: Insights into the application of omics in aquaculture settings under a microbiological perspective. Aquaculture. 738132. doi: 10.1016/j.aquaculture.2022.738132

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  CaoJ.XuL. (2018). “ Research on counting algorithm of residual feeds in aquaculture based on machine vision”, in: 2018 International Conference on Image, Vision and Computing (ICIVC), Piscataway, NJ: IEEE108–112. doi: 10.1109/ICIVC.2018.8492873

  • CrossRef
  • Google Scholar

• 25

  ChangC.-C.UbiñaN.ChengS.-C.LanH.-Y.ChenK.-C.HuangC.-C. (2022). “ A two-mode underwater smart sensor object for precision aquaculture based on AIoT technology”, in: Italian National Conference on Sensors ( MDPI), 22, 7603. doi: 10.3390/s22197603

  • CrossRef
  • Google Scholar

• 26

  ChengW. K.KhorJ. C.LiewW. Z.BeaK. T.ChenY.-L. (2024). Integration of federated learning and edge-cloud platform for precision aquaculture. IEEE Access12, 3454057. doi: 10.1109/ACCESS.2024.3454057

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  CordierT.EslingP.LejzerowiczF.ViscoJ.OuadahiA.MartinsC. I. M.et al. (2017). Predicting the ecological quality status of marine environments from eDNA metabarcoding data using supervised machine learning. Environ. Sci. Technol.51, 9118–9126. doi: 10.1021/acs.est.7b01518

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  CuiM.LiuX.LiuH.ZhaoJ.LiD.WangW. (2024). Fish tracking, counting, and behaviour analysis in digital aquaculture: A comprehensive review. arXiv. doi: 10.48550/arXiv.2406.17800

  • CrossRef
  • Google Scholar

• 29

  DahlR. E. (2017). A study on price volatility in the aquaculture market using value-at-risk (VaR). Aquaculture Econ Manage.21, 145–162. doi: 10.1080/13657305.2017.1262475

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30

  DasaklisT. K.VoutsinasT. G.TsoulfasG.CasinoF. (2022). A systematic literature review of blockchain-enabled supply chain traceability implementations. Sustainability14, 2439. doi: 10.3390/su14042439

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  DasariD.DongC.-D.SinghaniaR.TambatV.PiechotaG.PatelA. K. (2025). Harnessing microalgae for sustainable aquaculture and mariculture: Marine pollution mitigation and circular economy strategies. Mar. pollut. Bull. 118292. doi: 10.1016/j.marpolbul.2025.118292

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  DavidsonK.WhyteC.AleynikD.DaleA.GontarekS.KurekinA.et al. (2021). HABreports: Online early warning of harmful algal and biotoxin risk for the Scottish shellfish and finfish aquaculture industries. Front. Mar. Sci.8, 631732. doi: 10.3389/fmars.2021.631732

  • CrossRef
  • Google Scholar

• 33

  DaviesR.WalesA. (2019). Antimicrobial resistance on farms: A review including biosecurity and the potential role of disinfectants in resistance selection. Compr. Rev. Food Sci. Food Saf.18, 753–774. doi: 10.1111/1541-4337.12438

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34

  DavisK. (2014). Bridging the innovation-policy gap. SAIS Rev. Int. Affairs34, 1–15. doi: 10.1353/SAIS.2014.0015

  • CrossRef
  • Google Scholar

• 35

  DiEuliisD.LutesC. D.GiordanoJ. (2018). Biodata risks and synthetic biology: A critical juncture. J. Biosci. Bioengineering6, 159. doi: 10.4172/2157-2526.1000159

  • CrossRef
  • Google Scholar

• 36

  FernandesA.TurraE. M.AlvarengaÉ.R.PassafaroT.LopesF.AlvesG. F. O.et al. (2020). Deep learning image segmentation for extraction of fish body measurements and prediction of body weight and carcass traits in Nile tilapia. Comput. Electron. Agric.170, 105274. doi: 10.1016/j.compag.2020.105274

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37

  FitzgeraldA.IoannouC. C.ConsuegraS.DowseyA.Garcia de LeanizC. (2025). Machine vision applications for welfare monitoring in aquaculture: Challenges and opportunities. Aquacult. Fish Fish.5, e70036. doi: 10.1002/aff2.70036

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  Food and Agriculture Organization of the United Nations (2022). “ Scaling up inclusive digitalization in agricultural value chains APRC/22/6”, in: FAO Regional Conference for Asia and the Pacific, Thirty-sixth Session, Dhaka, Bangladesh, 8–11 March 2022. Rome: Food and Agriculture Organization of the United Nations. Available online at: http://www.fao.org (Accessed January 26, 2026).

  • Google Scholar

• 39

  FuS.TuJ.XiaB.LiX.LiuY. (2015). Dynamics and predictive modeling of Vibrio spp. in a recirculating aquaculture system for Atlantic salmon (Salmo salar L.). J. Fishery Sci. China22, 269–277.

  • Google Scholar

• 40

  FuadT. R.AhmedS.IvanS. (2025). AQUA20: A benchmark dataset for underwater species classification under challenging conditions. arXiv. doi: 10.48550/arXiv.2506.17455

  • CrossRef
  • Google Scholar

• 41

  FurtadoL. S.da SilvaT. L. C.FerreiraM. G. F.de MacedoJ. A. F.Cavalcanti MoreiraJ. K. (2023). A framework for digital transformation towards smart governance: Using big data tools to target SDGs in Ceará, Brazil. J. Urban Manage.12, 1–12. doi: 10.1016/j.jum.2023.01.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  GapparovA.FallahhuseinM.MatkarimovN.FernandesR. B.ChuponovS.SehgalR.et al. (2025). Blockchain-enabled supply chain traceability in sustainable aquatic farming. Int. J. Aquat. Res. Environ. Stud.5, 5–s1-07. doi: 10.70102/ijares/v5s1/5-s1-07

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43

  GargM.ParuiS. K.KandhariH.DavidR.LakshmiS.NavalP. (2025). “ Enabling smart farming with IoT and sensor networks: Case studies from developing countries”, in: 2025 International Conference on Automation and Computation (AUTOCOM), Piscataway, NJ: IEEE. 1–6. doi: 10.1109/AUTOCOM64127.2025.10956508

  • CrossRef
  • Google Scholar

• 44

  GoswamiM.SanilV.ChoudhryA.SrinivasanA.UdompanyawitC.DubrawskiA. (2023). AQuA: A benchmarking tool for label quality assessment. arXiv. doi: 10.48550/arXiv.2306.09467

  • CrossRef
  • Google Scholar

• 45

  GuritnoA. D.TanuputriM. R. (2017). Comparison of fish supply chain from aquaculture and sea catchment areas. J. Service Sci. Manage.10, 428–437. doi: 10.4236/JSSM.2017.104028

  • CrossRef
  • Google Scholar

• 46

  GuterresB.SbrissaK.MendesA.MeirelesL.NovoveskáL.VermeulenF.et al. (2023). “ HAB detection within aquaculture industry: A case study in the Atlantic area,” in 2023 IEEE 21st International Conference on Industrial Informatics (INDIN). Piscataway, NJ: IEEE. doi: 10.1109/INDIN51400.2023.10218124

  • CrossRef
  • Google Scholar

• 47

  HarfoucheA. L.JacobsonD. A.KainerD.RomeroJ. C.HarfoucheA. H.Scarascia MugnozzaG.et al. (2019). Accelerating climate resilient plant breeding by applying next-generation artificial intelligence. Trends in Biotechnology37(11), 1217–1235. doi: 10.1016/j.tibtech.2019.05.007

  • CrossRef
  • Google Scholar

• 48

  HishamZ. S. H.AyamonA.FadzullahN. L. A. M.RusllimN. A. M.ZohdiM. A. M.AzharM. A. A.et al. (2025). The future of food security: The role of blockchain technology in global aquaculture. Int. J. Perceptive Cogn. Computing11, 540. doi: 10.31436/ijpcc.v11i1.540

  • CrossRef
  • Google Scholar

• 49

  HockingsE. (2016). “ A critical examination of policy-developments in information governance and the biosciences,” in Information Governance in the Biosciences (Cham: Springer), 45–62. doi: 10.1007/978-3-319-33525-4_5

  • CrossRef
  • Google Scholar

• 50

  HollandS.HosnyA.NewmanS.JosephJ.ChmielinskiK. (2018). “ The dataset nutrition label: A framework to drive higher data quality standards,” in Data Protection and Privacy (London: Bloomsbury Publishing), 1–22. doi: 10.5040/9781509932771.ch-001

  • CrossRef
  • Google Scholar

• 51

  HolzingerA.FisterI.FisterI.KaulH.-P.AssengS. (2024). Human-centered AI in smart farming: Toward Agriculture 5.0. IEEE Access12, 3395532. doi: 10.1109/ACCESS.2024.3395532

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  HolzingerA.SarantiA.AngerschmidA.RetzlaffC.GronauerA.PejakovićV.et al. (2022). “ Digital transformation in smart farm and forest operations needs human-centered AI: Challenges and future directions,” in Proceedings of the Italian National Conference on Sensors, Basel: MDPI. 3043. doi: 10.3390/s22083043

  • CrossRef
  • Google Scholar

• 53

  HossainB.SunnyA. R.Nayeem GaziM. M. R.DasA. R.MohajonR.MiahA. T. H.et al. (2024). Advancing fish farming through deep learning: Applications, opportunities, challenges, and future directions. Pathfinder Res.2, 39. doi: 10.69937/pf.por.2.3.39

  • CrossRef
  • Google Scholar

• 54

  HoustonR.BeanT.MacqueenD.GundappaM. K.JinY.JenkinsT. L.et al. (2020). Harnessing genomics to fast-track genetic improvement in aquaculture. Nat. Rev. Genet.21, 389–409. doi: 10.1038/s41576-020-0227-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  HuangY.-P.KhabusiS. P. (2025). Artificial Intelligence of Things (AIoT) advances in aquaculture: A review. Processes13, 73. doi: 10.3390/pr13010073

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  IsleyR. (2022). Algorithmic bias and its implications: How to maintain ethics through AI governance2. doi: 10.21428/4b58ebd1.0e834dbb

  • CrossRef
  • Google Scholar

• 57

  IsmailW. S. (2024). Human-centric AI: Enhancing user experience through natural language interfaces. J. Wireless Mobile Networks Ubiquitous Computing Dependable Appl.15, 12. doi: 10.58346/jowua.2024.i1.012

  • CrossRef
  • Google Scholar

• 58

  JaiesI.ShahF.QadiriS. S. N.QayoomI.BhatB. A.DarS. A.et al. (2024). Immunological and molecular diagnostic techniques in fish health: Present and future prospectus. Mol. Biol. Rep. doi: 10.1007/s11033-024-09344-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59

  JongjaraunsukR.TaparhudeeW.SuwannasingP. (2024). Comparison of water quality prediction for red tilapia aquaculture in an outdoor recirculation system using deep learning and a hybrid model. Water16, 907. doi: 10.3390/w16060907

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  KalyaniY.VorsterL.WhettonR.CollierR. (2024). Application scenarios of digital twins for smart crop farming through cloud-fog-edge infrastructure. Future Internet16, 100. doi: 10.3390/fi16030100

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  KamaliS.WardV.Ricardez‐SandovalL. (2023). Closed-loop operation of a simulated recirculating aquaculture system with an integrated application of nonlinear model predictive control and moving horizon estimation. Comput. Electron. Agric.208, 107820. doi: 10.1016/j.compag.2023.107820

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62

  KerrD.Butler-HendersonK.SahamaT. (2017). “ Security, privacy, and ownership issues with the use of wearable health technologies,” in Health Information Privacy and Security (Hershey, PA: IGI Global), 123–145. doi: 10.4018/978-1-5225-1016-1.CH007

  • CrossRef
  • Google Scholar

• 63

  KordzadehN.GhasemaghaeiM. (2021). Algorithmic bias: Review, synthesis, and future research directions. Eur. J. Inf. Syst.30, 596–621. doi: 10.1080/0960085X.2021.1927212

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64

  KumarV.RoyS.BarmanD.KumarA. (2014). Immunoserological and molecular techniques used in fish disease diagnosis: A mini review. Int. J. Fisheries Aquat. Stud.1, 111–117. Available online at: http://www.fisheriesjournal.com.

  • Google Scholar

• 65

  LaporteC. Y.O'ConnorR. V. (2016). “ A multi-case study analysis of software process improvement in very small companies using ISO/IEC 29110,” in European Conference on Software Process Improvement, Cham: Springer25–36. doi: 10.1007/978-3-319-44817-6_3

  • CrossRef
  • Google Scholar

• 66

  LeeJ.LooJ.ChuahY.TangP.TanY. (2013). The use of vision in a sustainable aquaculture feeding system. Res. J. Appl. Sciences Eng. Technol.6, 3228–3233. doi: 10.19026/RJASET.6.3573

  • CrossRef
  • Google Scholar

• 67

  LiD.LiX.WangQ.HaoY. (2022). Advanced techniques for the intelligent diagnosis of fish diseases: A review. Animals12, 2938. doi: 10.3390/ani12212938

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 68

  LiD.DuZ.WangQ.WangJ.DuL. (2023). Recent advances in acoustic technology for aquaculture: A review. Rev. Aquacult.15, 1147–1165. doi: 10.1111/raq.12842

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 69

  LiaoQ. V.VarshneyK. R. (2021). Human-centered explainable AI (XAI): From algorithms to user experiences. arXiv. doi: 10.48550/arXiv.2110.10790

  • CrossRef
  • Google Scholar

• 70

  LieØ. (2008). Improving Farmed Fish Quality and Safety (Boca Raton, FL, USA: CRC Press). doi: 10.1201/9781439833001

  • CrossRef
  • Google Scholar

• 71

  LimS.AndersonE. (2016). “ Institutional barriers against innovation diffusion: From the perspective of digital health startups”, in: Proceedings of the Hawaii International Conference on System Sciences, IEEE Computer Society4147–4156. doi: 10.1109/HICSS.2016.415

  • CrossRef
  • Google Scholar

• 72

  LimborgM. T.AlberdiA.KodamaM.RoggenbuckM.KristiansenK.GilbertM. T. P. (2018). Applied hologenomics: feasibility and potential in aquaculture. Trends Biotechnol.36, 252–264. doi: 10.1016/j.tibtech.2017.12.006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73

  LinkovI.TrumpB. D.Poinsatte-JonesK.FlorinM.-V. (2018). Governance strategies for a sustainable digital world. Sustainability10, 440. doi: 10.3390/su10020440

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74

  LittleD.MackenzieS. (2023). Grand challenges for global aquaculture. Front. Aquaculture. 2, 1232936. doi: 10.3389/faquc.2023.1232936

  • CrossRef
  • Google Scholar

• 75

  LiuL.ChengW.KuoH.-W. (2025). A narrative review on smart sensors and IoT solutions for sustainable agriculture and aquaculture practices. Sustainability17, 5256. doi: 10.3390/su17125256

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 76

  LiuC.CuiR. (2024). “ 3D design software and system adoption to streamline supply chain management: A case study of innovators of LincTex Digital,” in Bridging the Divide. doi: 10.31274/itaa.17117

  • CrossRef
  • Google Scholar

• 77

  LuY.YangY.ZhaoQ.ZhangC.LiT. J. (2024). AI assistance for UX: A literature review through human-centered AI. arXiv. doi: 10.48550/arXiv.2402.06089

  • CrossRef
  • Google Scholar

• 78

  LvJ.DengS.ZhangL. (2020). A review of artificial intelligence applications for antimicrobial resistance. Biosaf. Health2, 211–219. doi: 10.1016/j.bsheal.2020.08.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 79

  MacaulayS.EllisonA.KilleP.CableJ. (2022). Moving towards improved surveillance and earlier diagnosis of aquatic pathogens: From traditional methods to emerging technologies. Rev. Aquacult. 14(4), 1813–1829. doi: 10.1111/raq.12674

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 80

  MaiT. T. N. (2010). Enhancing quality management of fresh fish supply chains through improved logistics and ensured traceability. Reykjavík: University of Iceland.

  • Google Scholar

• 81

  MairG.HalwartM.YuD.Costa‐PierceB. (2023). A decadal outlook for global aquaculture. J. World Aquacult Soc. 54(2), 196–205. doi: 10.1111/jwas.12977

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 82

  MantovaniA.FerrariD.FrazzoliC. (2015). Sustainability, security and safety in the feed-to-fish chain: Focus on toxic contamination. Int. J. Nutr. Food. Sci.4, 41–49. doi: 10.11648/J.IJNFS.S.2015040202.12

  • CrossRef
  • Google Scholar

• 83

  MarvinH.van AsseltE. V.KleterG.MeijerN.LorentzenG.JohansenL.et al. (2020). Expert-driven methodology to assess and predict the effects of drivers of change on vulnerabilities in a food supply chain: Aquaculture of Atlantic salmon in Norway as a showcase. Trends Food Sci. Technol.100, 87–101. doi: 10.1016/j.tifs.2020.06.022

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 84

  MilijasevicM.Vesković-MoračaninS.Babic MilijasevicJ.PetrovicJ.NastasijevićI. (2024). Antimicrobial resistance in aquaculture: Risk mitigation within the One Health context. Foods13, 2448. doi: 10.3390/foods13152448

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 85

  MiñónR.Díaz-de-ArcayaJ.Torre-BastidaA. I.HartliebP. (2022). Pangea: an MLOps tool for automatically generating infrastructure and deploying analytic pipelines in edge, fog and cloud layers. Sensors22, 4425. doi: 10.3390/s22124425

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 86

  MontesclarosJ. M. L.BabuS. C.TengP. S. (2019). IoT-enabled farms and climate-adaptive agriculture technologies: investment lessons from Singapore (IFPRI Discussion Paper 1805) (Washington, DC: IFPRI/SSRN). Available online at: https://ssrn.com/abstract=3363740 (Accessed January 26, 2026).

  • Google Scholar

• 87

  MustafaS.EstimA.ShapawiR.ShalehandM. J.SidikS. (2021). Technological applications and adaptations in aquaculture for progress towards sustainable development and seafood security. IOP Conf. Series: Earth Environ. Sci.718, 12041. doi: 10.1088/1755-1315/718/1/012041

  • CrossRef
  • Google Scholar

• 88

  NguyenN. H.SonessonA.HoustonR.MoghadamH. (2022). Editorial: applications of modern genetics and genomic technologies to enhance aquaculture breeding. Front. Genet.13, 898857. doi: 10.3389/fgene.2022.898857

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 89

  NishiguchiM.HabeH.AbeK.OtaniM.IguchiN. (2025). “ A computer vision approach to counting farmed fish in flowing water”, in: Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2025): VISAPP. 3912. doi: 10.5220/0013388800003912

  • CrossRef
  • Google Scholar

• 90

  ObirikorangK. A.QuagrainieK.KassahJ. E.Von AhnenM. (2024). Editorial: sustainable aquaculture production for improved food security. Front. Sustain. Food Syst.8, 1485956. doi: 10.3389/fsufs.2024.1485956

  • CrossRef
  • Google Scholar

• 91

  OverreinM. M.TinnP.AldridgeD.JohnsenG.FragosoG. (2024). Biomass estimations of cultivated kelp using underwater RGB images from a mini-ROV and computer vision approaches. Front. Mar. Sci.11, 1324075. doi: 10.3389/fmars.2024.1324075

  • CrossRef
  • Google Scholar

• 92

  ÖzM.ÜstünerE. (2026). Artificial intelligence for Blue Transformation: a review of predictive modeling and decision support systems in sustainable aquaculture. Sustain. Dev. doi: 10.1002/sd.71290

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 93

  ÖzM.ÜstünerE.ÇifciS.BudakF.İleriE.DikelS. (2025). Artificial intelligence for fish disease diagnosis and management: innovations, challenges, and One Health implications. Aquacult. Int.33, 1–36. doi: 10.1007/s10499-025-02352-7

  • CrossRef
  • Google Scholar

• 94

  PadmanabanR.SudalaimuthuK. (2012). Marine fishery information system and aquaculture site selection using remote sensing and GIS. Int. J. Advanced Remote Sens. GIS1, 20–33.

  • Google Scholar

• 95

  PajoJ. F.HaukøM.Skaret-ThoresenR.GonzalezA.LehneP.GrøndalenO. (2023). “ Digitalization in the aquaculture industry: validation trials over a commercial 5G network”, in: 2023 Joint European Conference on Networks and Communications & 6G Summit (EuCNC/6G Summit), IEEE266–271. doi: 10.1109/EuCNC/6GSummit58263.2023.10188310

  • CrossRef
  • Google Scholar

• 96

  ParaskevasK.KavallieratouE. (2022). Biofouling detection in fish farming nets using neural networks. Soc. Sci. Res. Network. doi: 10.2139/ssrn.4144161

  • CrossRef
  • Google Scholar

• 97

  PartelowS.AsifF.BénéC.BushS.ManlosaA. O.NagelB.et al. (2023). Aquaculture governance: five engagement arenas for sustainability transformation. Curr. Opin. Environ. Sustainability64, 101379. doi: 10.1016/j.cosust.2023.101379

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 98

  PatrónG. D.Ricardez-SandovalL. A. (2024). Economically optimal operation of recirculating aquaculture systems under uncertainty. Comput. Electron. Agric.220, 108856. doi: 10.1016/j.compag.2024.108856

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 99

  PendharkarK. (2023). Algorithmic bias and discrimination: legal and policy considerations. Soc. Sci. Res. Network. doi: 10.2139/ssrn.4640433

  • CrossRef
  • Google Scholar

• 100

  PetersL.SpatharisS.DarioM. A.DwyerT.RocaI. J. T.KintnerA. H.et al. (2018). Environmental DNA: a new low-cost monitoring tool for pathogens in salmonid aquaculture. Front. Microbiol.9, 3009. doi: 10.3389/fmicb.2018.03009

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 101

  Puig-BoschX.BoonstraM.CabritaM.PerramonJ.MuniveS.GualaA.et al. (2024). “ Requirements for human-centered artificial intelligence: a heart failure study across Europe and Latin America”, in: Proceedings of the Symposium on Medical Information Processing and Analysis (Piscataway, NJ: IEEE), 1–8. doi: 10.1109/SIPAIM62974.2024.10783548

  • CrossRef
  • Google Scholar

• 102

  QuansahJ.RochonG.QuagrainieK. (2007). “ Remote sensing applications for sustainable aquaculture in Africa”, in: 2007 IEEE International Geoscience and Remote Sensing Symposium (Piscataway, NJ: IEEE), 2074–2077. doi: 10.1109/IGARSS.2007.4423034

  • CrossRef
  • Google Scholar

• 103

  RadiartaI. N.SaitohS.YasuiH. (2011). Aquaculture site selection for Japanese kelp (Laminaria japonica) in southern Hokkaido, Japan, using satellite remote sensing and GIS-based models. ICES J. Mar. Sci.68, 1501–1512. doi: 10.1093/ICESJMS/FSQ163

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 104

  RagabS.HoseinifarS.Van DoanH.RossiW.DaviesS. J.AshourM.et al. (2025). Overview of aquaculture artificial intelligence (AAI) applications: enhance sustainability and productivity, reduce labor costs, and increase the quality of aquatic products. Ann. Anim. Sci. 25(2), 441–453. doi: 10.2478/aoas-2024-0075

  • CrossRef
  • Google Scholar

• 105

  RatherM. A.AhmadI.ShahA.HajamY. A.AminA.KhursheedS.et al. (2024). Exploring opportunities of artificial intelligence in aquaculture to meet increasing food demand. Food. Chemistry: X22, 101309. doi: 10.1016/j.fochx.2024.101309

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 106

  Redman-WhiteC. J.MoranD.PetersA.MuwongeA. (2023). A review of the predictors of antimicrobial use and resistance in European food animal production. Front. Antibiot.3, 1209552. doi: 10.3389/frabi.2023.1209552

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 107

  RillingS.NielsenM.MilellaA.JestelC.FröhlichP.ReinaG. (2017). “ A multisensor platform for comprehensive detection of crop status: results from two case studies”, in: 2017 IEEE Advanced Video and Signal Based Surveillance (AVSS), IEEE1–6. doi: 10.1109/AVSS.2017.8078479

  • CrossRef
  • Google Scholar

• 108

  RowanN. (2022). The role of digital technologies in supporting and improving fishery and aquaculture across the supply chain – Quo Vadis? Aquacult. Fish.7, 233–246. doi: 10.1016/j.aaf.2022.06.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 109

  SaberioonM.GholizadehA.CísarP.PautsinaA.UrbanJ. (2017). Application of machine vision systems in aquaculture with emphasis on fish: state-of-the-art and key issues. Rev. Aquacult.9, 425–436. doi: 10.1111/RAQ.12143

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 110

  SaupharC.StolzC.TueneS.GanselL.AasG. H. (2023). Atlantic salmon (Salmo salar) distribution and vertical size-stratification in a commercial sea cage: a case study. Aquaculture569, 740356. doi: 10.1016/j.aquaculture.2023.740356

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 111

  SevinS.DikelS. (2025). Building a Collaborative Aquaculture Research Ecosystem with APIs and AI. Aquat. Sci. Eng.40 (1), 42–52. doi: 10.26650/ASE20241564766

  • CrossRef
  • Google Scholar

• 112

  SchiffD. S.BiddleJ.BorensteinJ.LaasK. (2019). “ What's next for AI ethics, policy, and governance? a global overview”, in: Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, ACM1–8. doi: 10.1145/3375627.3375804

  • CrossRef
  • Google Scholar

• 113

  ShaginimolC. N.ManojkumarB.KalirajS. (2023). Application of remote sensing and GIS in site selection for sustainable aquaculture in India: a review. pollut. Res.42, 1. doi: 10.53550/pr.2023.v42i02.009

  • CrossRef
  • Google Scholar

• 114

  ShamsuzzohaA.MarttilaJ.HeloP. (2023). Blockchain-enabled traceability system for the sustainable seafood industry. Technol. Anal. Strategic Manage.35, 873–888. doi: 10.1080/09537325.2023.2233632

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 115

  SharmaD.KumarR. (2021). “ Smart aquaculture: integration of sensors, biosensors, and artificial intelligence,” in Concepts and Strategies in Plant Sciences (Cham: Springer), 341–358. doi: 10.1007/978-3-030-66165-6_21

  • CrossRef
  • Google Scholar

• 116

  ShneidermanB. (2020). Human-centered artificial intelligence: reliable, safe & trustworthy. Int. J. Hum.-Comput. Interact.36, 495–504. doi: 10.1080/10447318.2020.1741118

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 117

  ShreeshaS.PaiM. M. M.PaiR. M.VermaU. (2023). Pattern detection and prediction using deep learning for intelligent decision support to identify fish behaviour in aquaculture. Ecological Informatics78, 102287. doi: 10.1016/j.ecoinf.2023.102287

  • CrossRef
  • Google Scholar

• 118

  SkaldebøM.SchellewaldC.EvjemoL.AmundsenH. B.XanthidisM.KelasidiE. (2024). “ Approaches enabling underwater autonomy and sensing in sea-based aquaculture settings”, in: 2024 32nd Mediterranean Conference on Control and Automation (MED) (Piscataway, NJ: IEEE), 1–6. doi: 10.1109/MED61351.2024.10566210

  • CrossRef
  • Google Scholar

• 119

  SmithM. E.BernardS. (2020). Satellite ocean color based harmful algal bloom indicators for aquaculture decision support in the Southern Benguela. Front. Mar. Sci.7, 61. doi: 10.3389/fmars.2020.00061

  • CrossRef
  • Google Scholar

• 120

  SnyderJ. N.BossE.WeatherbeeR. A.ThomasA.BradyD.NewellC. (2017). Oyster aquaculture site selection using Landsat 8-derived sea surface temperature, turbidity, and chlorophyll a. Front. Mar. Sci.4, 190. doi: 10.3389/fmars.2017.00190

  • CrossRef
  • Google Scholar

• 121

  SouzaR.MaltaK.SilvaR. S.MasieroP.AlmeidaE.MaChadoI. (2019). “ A case study about startups' software development practices: a preliminary result,” in Proceedings of the Brazilian Symposium on Software Quality, New York, NY: Association for Computing Machinery (ACM). 1–7. doi: 10.1145/3364641.3364663

  • CrossRef
  • Google Scholar

• 122

  SuX.SutarlieL.LohX. (2020). Sensors, biosensors, and analytical technologies for aquaculture water quality. Research2020, 8272705. doi: 10.34133/2020/8272705

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 123

  SubramaniT.Venkatachala MoorthyK. K.PriyankaS. (2017). Assessment of impact on aquaculture using remote sensing data and GIS in Tiruchendur. Int. J. Emerging Trends Technol. Comput. Sci. (IJETTCS)6, 157–166.

  • Google Scholar

• 124

  SunM.YangX.XieY. (2020). Deep learning in aquaculture: a review. J. Comput.31, 294–319. doi: 10.3966/199115992020023101028

  • CrossRef
  • Google Scholar

• 125

  SundarayJ.DixitS.RatherA.RasalK.SahooL. (2022). Aquaculture omics: an update on the current status of research and data analysis. Mar. Geonomics. 64, 100967. doi: 10.1016/j.margen.2022.100967

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 126

  SwethaP.RasheedA.HarigovindanV. P. (2023). “ Random forest regression based water quality prediction for smart aquaculture”, in: 2023 4th International Conference on Computing and Communication Systems (I3CS) (Piscataway, NJ: IEEE), 1–6. doi: 10.1109/I3CS58314.2023.10127488

  • CrossRef
  • Google Scholar

• 127

  TaconA.MetianM. (2008). Aquaculture feed and food safety. Ann. N. Y. Acad. Sci.1140, 50–57. doi: 10.1196/annals.1454.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 128

  TaylorK.GriffithC.LefortL.GaireR.ComptonM.WarkT.et al. (2013). Farming the web of things. IEEE Intell. Syst.28, 72–80. doi: 10.1109/MIS.2013.102

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 129

  ThaiT. T. N.NguyenT. S.PhamV. (2021). “ Computer vision based estimation of shrimp population density and size”, in: 2021 International Symposium on Electrical and Electronics Engineering (ISEE), IEEE90–94. doi: 10.1109/ISEE51682.2021.9418638

  • CrossRef
  • Google Scholar

• 130

  ThakurK.ShettyM.SinghS.KhanapuriJ. (2023). “ Enhancing fish disease detection: A comprehensive review for sustainable aquaculture”, in: 2023 International Conference on Awareness Science and Technology (iCAST) (Piscataway, NJ: IEEE), 77–82. doi: 10.1109/ICAST59062.2023.10454980

  • CrossRef
  • Google Scholar

• 131

  TroellM.Costa‐PierceB.SteadS.CottrellR. S.BrugereC.FarmeryA.et al. (2023). Perspectives on aquaculture's contribution to the Sustainable Development Goals for improved human and planetary health. J. World Aquacult Soc. 54(2), 251–342. doi: 10.1111/jwas.12946

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 132

  UbiñaN.LanH.-Y.ChengS.-C.ChangC.-C.LinS.-S.ZhangK.et al. (2023). Digital twin-based intelligent fish farming with artificial intelligence Internet of Things (AIoT) ( Social Science Research Network). doi: 10.2139/ssrn.4379114

  • CrossRef
  • Google Scholar

• 133

  WangX.WuY.JiX.FuH. (2024). Algorithmic discrimination: Examining its types and regulatory measures with emphasis on US legal practices. Front. Artif. Intell.7, 1320277. doi: 10.3389/frai.2024.1320277

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 134

  WilkinsonM. D.VerborghR.Bonino da Silva SantosL. O.ClarkT.SwertzM.KelpinF.et al. (2017). Interoperability and FAIRness through a novel combination of web technologies. PeerJ Comput. Sci.3, e110. doi: 10.7717/peerj-cs.110

  • CrossRef
  • Google Scholar

• 135

  WrightA.LiX.YangX.SotoE.GrossJ. A. (2023). Disease prevention and mitigation in US finfish aquaculture: A review of current approaches and new strategies. Rev. Aquacult.15, 2159–2179. doi: 10.1111/raq.12807

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 136

  WuA.-Q.LiK.-L.SongZ.-Y.LouX.HuP.YangW.et al. (2025). Deep learning for sustainable aquaculture: Opportunities and challenges. Sustainability17, 5084. doi: 10.3390/su17115084

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 137

  XiaoX.FuZ.ZhangY.PengZ.XiaoshuanZ. (2016). Developing an intelligent traceability system for aquatic products in cold chain logistics integrated WSN with SPC. J. Food Process. Preserv.40, 12730. doi: 10.1111/JFPP.12730

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 138

  YoungA.ZahuranecA.WinowatanM.VerhulstS. (2019). Leveraging private data for public good: A descriptive analysis and typology of existing practices. Foundation Rev.11, 1–15. doi: 10.15868/socialsector.40380

  • CrossRef
  • Google Scholar

• 139

  YuR.LeungP. (2006). Optimal Partial Harvesting Schedule for Aquaculture Operations. Marine Resource Economics, 21(3), 301–315. doi: 10.1086/mre.21.3.42629513

  • CrossRef
  • Google Scholar

• 140

  YuR.LeungP.KamL. E.BienfangP. (2010). “ A decision support system for scheduling partial harvesting in aquaculture,” in Decision Support Systems in Agriculture, Food and the Environment: Trends, Applications and Advances (Hershey, PA: IGI Global), 309–324. doi: 10.4018/978-1-61520-881-4.ch018

  • CrossRef
  • Google Scholar

• 141

  ZambranoA.GiraldoL. F.QuimbayoJ.MedinaB.CastilloE. (2021). Machine learning for manually-measured water quality prediction in fish farming. PloS One16, e0256380. doi: 10.1371/journal.pone.0256380

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 142

  ZhangY.AtiqA.ChowW. (2024). Exploring the role of AI in UX research. ASCILITE Publications. 1341. doi: 10.14742/apubs.2024.1341

  • CrossRef
  • Google Scholar

• 143

  ZhaoY.QinH.XuL.YuH.ChenY. (2025). A review of deep learning-based stereo vision techniques for phenotype feature and behavioral analysis of fish in aquaculture. Artificial Intelligence Review58: 7. doi: 10.1007/s10462-024-10960-7

  • CrossRef
  • Google Scholar