How transparent is water? Addressing ethical concerns in AI surrogate development for nearshore wave modelling and the natural sciences

Purpose This paper aims to propose an ethical framework to guide artificial intelligence (AI) developers within natural sciences and engineering, to prioritise responsible and context-sensitive model development. Design/methodology/approach This paper tackles the themes of bias and transparency, which are common in AI discourse. These themes are addressed with a focus on developing AI models to understand offshore wave conditions, a growing field in AI development. The paper goes on to provide a development framework that can be adapted to any AI development within engineering and physical sciences. Findings Transparency and bias in AI is a subjective construct, and AI is made for an intended population of end-users. Prioritising the expected domain knowledge and collaborative approach to AI development in engineering and physical sciences, AI in these fields can be responsibly adopted in everyday workflows. Research limitations/implications This research addresses ethical concerns within the context of nearshore wave modelling, specifically targeting the issues of transparency and bias through the lens of scientific precision. As a discussion paper, it lays the groundwork for applying similar methodologies across other areas of the physical sciences and engineering. However, for such adaptation, ethical considerations must be tailored to the specific characteristics of the physical phenomena being modelled. Practical implications The paper is intended to provide perspective development of AI surrogates in the physical sciences, using nearshore wave modelling as an example. The proposed circular workflow is presented to be adopted in AI development for surrogate models of natural systems. Social implications This paper highlights the social implications of ethically developing AI for nearshore wave modelling, a critical tool in offshore construction. As societies push for rapid expansion of renewable energy and offshore infrastructure, AI plays a key role in meeting global development goals. Yet, without ethical safeguards, these technologies risk amplifying existing vulnerabilities. The framework proposed here promotes socially responsible AI development and can be extended to other high-stakes applications in engineering and the physical sciences. Originality/value There is a lack of published work that address ethical questions in AI within a technical setting. The contribution of this work is to help formalise a logically consistent approach towards ethical consideration of AI consistent with popularised AI discourse centred on natural language processing and facial recognition.