Luo et al. 2026 carry the agentic-science paradigm into atmosphere–ocean science: their EarthLink copilot interprets climate researchers' intent, generates experimental plans, and writes analysis code, and they envision autonomous, goal-oriented Earth-science agents that design experiments and test hypotheses from multisource data.
The cited text presents EarthLink as an externally cited system ([278]) that 'represents a step in this direction,' not as Luo et al.'s own copilot, so 'their EarthLink copilot' misattributes ownership; the functional description of EarthLink and the envisioning of autonomous, goal-oriented Earth-science agents are otherwise supported.
Written by Kimi K3 via Ollama Cloud · checked by GLM-5.3 via Ollama Cloud · 6 Oct, 04:37
Source chain
AEvery quote below was checked, without a model, to appear verbatim in its source.
- 01
“EarthLink—an AI agent system designed as an interactive copilot for climate scientists”
Full-text passage · no page number · evidence E27
These advances highlight a broader trend toward the emergence of the ‘AI scientist’, in which machines evolve from analytical tools into collaborative partners that support and extend human intellect. Earth science is beginning to follow this trajectory. The central challenge in this domain is not simply to improve prediction from ever larger datasets, but to achieve a deeper, human-like understanding of the physical world, especially when working with sparse and heterogeneous observations. Meeting this challenge requires moving beyond pattern recognition toward genuine scientific reasoning. A step in this direction is represented by EarthLink—an AI agent system designed as an interactive copilot for climate scientists [278]. Researchers can pose questions in natural language and EarthLink interprets intent, generates experimental plans, writes analysis code and synthesizes results into transparent and traceable research reports, such as evaluating the annual precipitation cycle in Asia (Fig. 9). By learning from successful cases through a practice–learning–improvement cycle, EarthLink continuously evolves to address increasingly complex problems.
Passage read from www.ebi.ac.uk, which may be a preprint rather than the published version.
- 02
“interprets intent, generates experimental plans, writes analysis code and synthesizes results into transparent and traceable research reports”
Full-text passage · no page number · evidence E27
These advances highlight a broader trend toward the emergence of the ‘AI scientist’, in which machines evolve from analytical tools into collaborative partners that support and extend human intellect. Earth science is beginning to follow this trajectory. The central challenge in this domain is not simply to improve prediction from ever larger datasets, but to achieve a deeper, human-like understanding of the physical world, especially when working with sparse and heterogeneous observations. Meeting this challenge requires moving beyond pattern recognition toward genuine scientific reasoning. A step in this direction is represented by EarthLink—an AI agent system designed as an interactive copilot for climate scientists [278]. Researchers can pose questions in natural language and EarthLink interprets intent, generates experimental plans, writes analysis code and synthesizes results into transparent and traceable research reports, such as evaluating the annual precipitation cycle in Asia (Fig. 9). By learning from successful cases through a practice–learning–improvement cycle, EarthLink continuously evolves to address increasingly complex problems.
Passage read from www.ebi.ac.uk, which may be a preprint rather than the published version.
- 03
“autonomous, goal-oriented systems capable of designing and running experiments, generating and testing hypotheses, and learning dynamics from multisource data”
Full-text passage · no page number · evidence E26
A new framework for AI-based model intercomparison is essential for rigorous benchmark performance. Finally, we contextualize these technical developments by discussing the usefulness and applicability of AI to society, including the improvement of multi-hazard early-warning systems and green energy production. We conclude by envisioning the future of AI agents for Earth science—autonomous, goal-oriented systems capable of designing and running experiments, generating and testing hypotheses, and learning dynamics from multisource data. This synthesis underscores that AI is not merely a tool, but a paradigm shift, which will significantly improve how we understand and adapt to a changing climate. INTRODUCTION Atmospheric and oceanic sciences have progressed through the combined supports of monitoring, theoretical analysis and numerical modeling over the past century. Intensified observations reveal major atmosphere and ocean phenomena—frontal systems, eddies, waves and large‑scale circulations—and make it possible to trace their complete life cycles.
Passage read from www.ebi.ac.uk, which may be a preprint rather than the published version.