The long-term value of artificial intelligence (AI) for water sciences remains the subject of debate. Here we argue that the core ideas underlying AI (rather than specific algorithms) can be integrated with domain knowledge to open new modes of scientific inquiry. We identify eight such core ideas, each offering capabilities difficult to achieve within traditional frameworks. Among these, expressive inductive bias allows architectures to declare what matters most while retaining capacity to learn the unexpected; differentiable programming enables constrained learning of partially known relationships; generative models capture joint distributions supporting conditional inference from fragmented observations; functional mapping learns solution operators providing orders-of-magnitude speedups for simulations; and inference-time computation enables AI to decompose complex problems. These ideas build on AI’s ability to learn useful representations of complex systems. Across many of them, a central mode of inquiry is ‘constrained learning’: posing unknown relationships as learnable components within known physical or conceptual frameworks. Tools built on these ideas can broaden access to reliable water predictions and support important open questions such as land–atmosphere interactions, organizing principles of the waterscape, groundwater sustainability, and human water-use behaviours. Domain insights can also feed back to advance general AI, benefiting the broader sciences. This Perspective argues that AI’s lasting value for water science lies in a number of underlying core ideas, which can be combined with domain knowledge to open new modes of inquiry into the global water cycle.
Methane accumulation in oxygenated freshwaters, known as the methane paradox, remains a poorly constrained component of aquatic methane cycling. Although methylphosphonate (MPn) hydrolysis is considered an important pathway for oxic methane production in marine systems, its relevance to freshwaters remains uncertain. Here we identify a rapid oxic methane production process in a seasonally stratified reservoir that occurs within hours and is stimulated by locally derived dissolved organic phosphorus (DOP). Cross-depth incubations showed that DOP-enriched filtrates enhanced methane accumulation, whereas inorganic phosphorus suppressed this response. In contrast, MPn amendments enhanced methane accumulation relative to unamended controls only after a pronounced lag, whereas phosphorus-31 nuclear magnetic resonance (31P-NMR) spectroscopy and metagenomic analyses provided no evidence for abundant MPn or canonical MPn biosynthesis. Together, these results indicate that rapid freshwater oxic methane production is associated with a labile, non-MPn DOP pool, expanding the prevailing MPn-centred view of oxic methane production. Oxic methane production in a freshwater reservoir was rapidly stimulated by a labile dissolved organic phosphorus pool rather than methylphosphonate hydrolysis, indicating an alternative mechanism underlying the methane paradox.
The ongoing conflict between India and Pakistan, heightened by the April 2025 terrorist attack in Pahalgam, Kashmir, has led to an unprecedented abeyance of India’s participation in the Indus Waters Treaty (IWT). Although this treaty has often been described as an example of hydrological cooperation despite regional conflict, it has systemic problems from the perspective of long-term environmental, political and social sustainability. We identify several factors that justify renegotiating and revising the IWT, leveraging multilateral and domestic policy frameworks to formulate a more effective and comprehensive agreement. This includes metrics set forth on transboundary water cooperation under the Sustainable Development Goals (Indicator 6.5.2) as well as a range of international legal guidelines for drafting ecologically viable treaties. We explore the role that nascent regional organizations could play in reducing tensions over the Indus Basin and facilitating new discussions. Ultimately, a reset in the IWT could catalyse broader confidence-building and a lasting peace in a region where two nuclear rivals share a 3,200-km-long contiguous border. The ongoing conflict between India and Pakistan, exacerbated by the April 2025 terrorist attack in Kashmir, has led to India’s unprecedented suspension of the Indus Waters Treaty. A renegotiation of the treaty should leverage international and domestic frameworks to enhance environmental, political and social sustainability, potentially fostering regional peace.
The latest directive from the European Union (EU) on the treatment of urban wastewater brings in stricter regulations intended to drive progress towards the United Nations’ sustainability goals, but also heightens the barriers to EU accession for many candidate countries. However, with the right support, this compliance challenge can be reframed as an opportunity to accelerate progress and strengthen environmental governance, not only for accession candidates but also for the existing EU and its neighbours.
Water bankruptcy is being promoted as a framing for escalating water challenges. Yet, it risks obscuring their political roots and reproducing technocratic approaches that leave underlying inequalities and drivers of overuse and depletion unaddressed. A transformative vision for global water action must confront the political economy of depletion, and centre justice, care, and plural ways of knowing water.
