GPT, large language models (LLMs) and generative artificial intelligence (GAI) models in geospatial science: a systematic review

International Journal of Digital Earth, Journal Year: 2024, Volume and Issue: 17(1)

Published: May 20, 2024

The launch of large language models (LLMs) like ChatGPT in late 2022 and the anticipated arrival future GPT-x iterations have marked beginning generative artificial intelligence (GAI) era. We conducted a systematic review how to integrate LLMs including GPT other GAI into geospatial science, based on 293 papers obtained from four databases academic publications – Web Science (WoS), Scopus, SSRN arXiv 26 were eventually included for analysis. statistically outlined share domains where models, type data that been used these modelling tasks roles they play. also pointed out challenges directions next research agenda along with which we could better position ourselves mainstream science cutting-edge paradigm as others leverage insights growing deluge.

Language: Английский

---

A synergistic future for AI and ecology

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(38)

Published: Sept. 11, 2023

Research in both ecology and AI strives for predictive understanding of complex systems, where nonlinearities arise from multidimensional interactions feedbacks across multiple scales. After a century independent, asynchronous advances computational ecological research, we foresee critical need intentional synergy to meet current societal challenges against the backdrop global change. These include unpredictability systems-level phenomena resilience dynamics on rapidly changing planet. Here, spotlight promise urgency convergence research paradigm between AI. Ecological systems are challenge fully holistically model, even using most prominent technique today: deep neural networks. Moreover, have emergent resilient behaviors that may inspire new, robust architectures methodologies. We share examples how modeling would benefit techniques themselves inspired by they seek model. Both fields each other, albeit indirectly, an evolution toward this convergence. emphasize more purposeful accelerate whilst building currently lacking modern which been shown fail at times because poor generalization different contexts. Persistent epistemic barriers attention disciplines. The implications successful go beyond advancing disciplines or achieving artificial general intelligence-they persisting thriving uncertain future.

Language: Английский

---

Detecting, attributing, and projecting global marine ecosystem and fisheries change: FishMIP 2.0

Authorea (Authorea), Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 22, 2024

There is an urgent need for models that can robustly detect past and project future ecosystem changes risks to the services they provide people. The Fisheries Marine Ecosystem Model Intercomparison Project (FishMIP) was established develop model ensembles projecting long-term impacts of climate change on fisheries marine ecosystems while informing policy at spatio-temporal scales relevant Inter-Sectoral Impact (ISIMIP) framework. While contributing FishMIP have improved over time, large uncertainties in projections remain, particularly coastal shelf seas where most world’s occur. Furthermore, previous impact mostly ignored fishing activity due a lack standardized historical scenario-based human forcing uneven capabilities dynamically across community. This, addition underrepresentation processes, has limited ability evaluate ensemble’s adequately capture states - crucial step building confidence projections. To address these issues, we developed two parallel simulation experiments (FishMIP 2.0) on: 1) evaluation detection 2) scenarios Key advances include forcing, captures oceanographic features not previously resolved, systematically test effects models. 2.0 key towards attribution framework regional global scales, enhanced relevance through increased ensemble

Language: Английский

---

Harnessing large language models for coding, teaching and inclusion to empower research in ecology and evolution

Methods in Ecology and Evolution, Journal Year: 2024, Volume and Issue: 15(10), P. 1757 - 1763

Published: May 2, 2024

Abstract Large language models (LLMs) are a type of artificial intelligence (AI) that can perform various natural processing tasks. The adoption LLMs has become increasingly prominent in scientific writing and analyses because the availability free applications such as ChatGPT. This increased use not only raises concerns about academic integrity but also presents opportunities for research community. Here we focus on using coding ecology evolution. We discuss how be used to generate, explain, comment, translate, debug, optimise test code. highlight importance effective prompts carefully evaluating outputs LLMs. In addition, draft possible road map inclusively with integrity. accelerate process, especially unfamiliar tasks, up time higher level tasks creative thinking while increasing efficiency output. enhance inclusion by accommodating individuals without skills, limited access education coding, or whom English is their primary written spoken language. However, code generated variable quality issues related mathematics, logic, non‐reproducibility intellectual property; it include mistakes approximations, novel methods. benefits teach learn advocate guiding students appropriate AI tools coding. Despite ability assign many LLMs, reaffirm continued teaching skills interpreting LLM‐generated develop critical skills. As editors MEE, support—to extent—the transparent, accountable acknowledged other publications. If comparable (excluding commonly aids like spell‐checkers, Grammarly Writefull) produce work described manuscript, there must clear statement effect its Methods section, corresponding senior author take responsibility any (or text) platform.

Language: Английский

---

Generative AI as a tool to accelerate the field of ecology

Nature Ecology & Evolution, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 29, 2025

Language: Английский

---

Interactions between Plant Communities and Water Environments in the Artificial Mangroves

Wetlands, Journal Year: 2025, Volume and Issue: 45(2)

Published: Feb. 1, 2025

Language: Английский

---

RARPKB: a knowledge-guide decision support platform for personalized robot-assisted surgery in prostate cancer

International Journal of Surgery, Journal Year: 2024, Volume and Issue: unknown

Published: March 18, 2024

Background: Robot-assisted radical prostatectomy (RARP) has emerged as a pivotal surgical intervention for the treatment of prostate cancer. However, complexity clinical cases, heterogeneity cancer, and limitations in physician expertise pose challenges to rational decision-making RARP. To address these challenges, we aimed organize knowledge previously complex cohorts establish an online platform named RARP Knowledge Base (RARPKB) provide reference evidence personalized plans. Materials Methods: PubMed searches over past two decades were conducted identify publications describing We collected, classified, structured details, patient information, data, various statistical results from literature. A knowledge-guided decision-support tool was established using MySQL, DataTable, ECharts, JavaScript. ChatGPT-4 assessment scales used validate compare platform. Results: The comprised 583 studies, 1589 cohorts, 1 911 968 patients, 11 986 records, resulting 54 834 data entries. decision support plan recommendations potential complications on basis patients’ baseline information. Compared with ChatGPT-4, RARPKB outperformed authenticity (100% versus [vs.] 73%), matching vs. 53%), 20%), patients 0%), 20%). Post-use, average System Usability Scale score 88.88±15.03, Net Promoter Score 85. base is available at http://rarpkb.bioinf.org.cn. Conclusions: introduced pioneering RARPKB, first robot-assisted surgery, emphasis can assist planning cancer improve its efficacy. provides future applications artificial intelligence practice.

Language: Английский

---

Detecting, Attributing, and Projecting Global Marine Ecosystem and Fisheries Change: FishMIP 2.0

Earth s Future, Journal Year: 2024, Volume and Issue: 12(12)

Published: Dec. 1, 2024

Abstract There is an urgent need for models that can robustly detect past and project future ecosystem changes risks to the services they provide people. The Fisheries Marine Ecosystem Model Intercomparison Project (FishMIP) was established develop model ensembles projecting long‐term impacts of climate change on fisheries marine ecosystems while informing policy at spatio‐temporal scales relevant Inter‐Sectoral Impact (ISIMIP) framework. While contributing FishMIP have improved over time, large uncertainties in projections remain, particularly coastal shelf seas where most world's occur. Furthermore, previous impact been limited by a lack global standardized historical fishing data, low resolution processes, uneven capabilities across community dynamically fisheries. These features are needed evaluate how reliably ensemble captures states ‐ crucial step building confidence projections. To address these issues, we developed 2.0 comprising two‐track framework for: (a) evaluation attribution (b) socioeconomic scenario Key advances include forcing, which oceanographic not previously resolved, forcing test effects systematically models. toward detection changing enhanced relevance through increased Our results will help elucidate pathways achieving sustainable development goals.

Language: Английский

---

Harnessing Large Language Models for Coding, Teaching, and Inclusion to Empower Research in Ecology and Evolution

Published: Feb. 28, 2024

1. Large language models (LLMs) are a type of artificial intelligence (AI) that can perform various natural processing tasks. The adoption LLMs has become increasingly prominent in scientific writing and analyses because the availability free applications such as ChatGPT. This increased use raises concerns about academic integrity, but also presents opportunities for research community. Here we focus on using coding ecology evolution. We discuss how be used to generate, explain, comment, translate, debug, optimise, test code. highlight importance effective prompts carefully evaluating outputs LLMs. In addition, draft possible road map inclusively with integrity.2. accelerate process, especially unfamiliar tasks, up time higher-level tasks creative thinking while increasing efficiency output. enhance inclusion by accommodating individuals without skills, limited access education coding, or whom English is not their primary written spoken language. However, code generated variable quality issues related mathematics, logic, non-reproducibility, intellectual property; they include mistakes approximations, novel methods.3. benefits teach learn advocate guiding students appropriate AI tools coding. Despite ability assign many LLMs, reaffirm continued teaching skills interpreting LLM develop critical skills.4. As editors MEE, support—to extent—the transparent, accountable, acknowledged other publications. If comparable (excluding commonly-used aids like spell-checkers, Grammarly Writefull) produce work described manuscript, there must clear statement effect its Methods section, corresponding senior author take responsibility any (or text) platform.

Language: Английский

---

Evaluating the ability of convolutional neural networks for transfer learning in Pinus radiata cover predictions

Ecological Informatics, Journal Year: 2024, Volume and Issue: 82, P. 102684 - 102684

Published: June 13, 2024

The species Pinus radiata is highly invasive in native forests Chile, drastically affecting the functioning and structure of ecosystems. Hence, it imperative to develop robust approaches detect P. invasions at different scales. Models based on convolutional neural networks (CNN) have proven be a promising alternative plant high-resolution remote sensing data, such as those obtained by drones. However, studies been limited their spatial variability assessments transferability or transfer learning new sectors, hindering ability use these models real-world setting. We train CNN architectures using unpiloted aerial vehicle data evaluate outside training domain regression approaches. compared trained with low (mono-site) high (multi-site). further sought maximize transference searching among models, maximizing evaluation an independent set. results showed that better when multi-site higher are used for training, obtaining coefficient determination R2 between 60% 87%. On contrary, mono-site present wide performance attributed dissimilarity information sites, limiting possibilities extrapolations model generalizations. also significant difference within-domain generalization test versus domain, showing testing alone cannot depict discrepancy without data. Finally, best domains often do not agree selected standard training/validation/testing scheme. Our findings pave way deeper discussions investigations into limitations applied imagery.

Language: Английский

---