Is Artificial Intelligent Governance Smart? Assessing AI in Marine Management Use

In spite of being essential to the functioning of physical, biological and societal processes, the oceans are often ‘out of sight and mind’. However, technological advances including remote sensing techniques, ocean sensor networks and underwater robots enable the acquisition and analysis of marine big data making oceans somehow known. Through the use of AI, more and more ‘smart’ ocean technologies such as intelligent ocean sensor networks are emerging, marking a new era in ocean observation, identification of ocean processes and ocean forecasting (1). As AI can bridge and fill the gaps in ocean data and enhance ocean modelling, it creates virtual representations of the ocean, called Digital Twins of the Ocean. The development of the European Digital Twin Ocean is believed to be of crucial importance in reaching the EU mission ‘Restore our Oceans and Waters’ by 2030 as “it will transform data into knowledge for everyone's benefit” and “will help design the most effective ways to restore marine and coastal habitats (...) and mitigate and adapt to climate change” (2). Likewise, the 2024 ‘AI for Good’ impact report states that AI could be a key driver in “getting back on track” to meet the United Nations Sustainable Development Goals (including SDG 14 ‘Life Below Water’) “without leaving anyone behind” (3). However, historically it has been shown that advancements in technology lead first and foremost to increasing extraction of marine living and non-living resources resulting in overfishing and fossil fuel-driven climate change. Hence, this paper raises critical questions to the belief that AI will play a main role in solving the ‘grand challenges of our time’ towards reaching SDG 14. Despite the potential of AI in optimising certain processes (e.g. the identification of fish), AI cannot compensate for poor marine management decisions. On the contrary, the application of AI faces a multitude of risks with potential negative social consequences, raising the question by and for whom AI ‘solutions’ are developed and deployed. The presentation will elaborate on how biases and flaws in AI training data (e.g. due to lacking integration of diverse knowledges), narrow optimization of AI and deployment risks raise ethical concerns and may further exacerbate environmental injustice and skew power dynamics in and around the marine realm.

(1)Song et al. (2023) A review of artificial intelligence in marine science, Frontiers in Earth Science, https://doi.org/10.3389/feart.2023.1090185

(2) EU official webpage, retrieved on 06.11.2024, https://research-and-innovation.ec.europa.eu/ funding/funding-opportunities/funding-programmes-and-open-calls/horizon-europe/eu-missions-horizon-europe/restore-our-ocean-and-waters/european-digital-twin-ocean-european-dto_en

(3) AI for Good Impact Report (2024) International Telecommunication Union, Geneva, Switzerland, retrieved on 06.11.2024 https://aiforgood.itu.int/newsroom/publications-and-reports/