Nature
A leading international scientific journal covering all fields of the natural sciences and publishing frontier research in areas such as materials science and AI.
Core Facts
A 2016 survey by Nature showed that more than 70% of researchers had attempted to replicate others' experiments and failed.
模型坍缩（Model Collapse）概念在2023年由牛津大学和剑桥大学研究团队在Nature论文中正式提出
In early 2023, top journals such as Science and Nature explicitly prohibited listing AI as a paper author, but allowed disclosure of AI's assistive use in the methodology section.
Timeline (last 90 days)
GNoME is a project published by Google DeepMind in Nature in 2023 that uses graph neural networks to predict the stability of inorganic crystal structures, producing about 2.2 million new crystal structures at once, of which about 380,000 were assessed as stable candidates.
Google 于 2021 年在 Nature 上发表了使用强化学习优化芯片布局布线（chip floorplanning）的研究
The core technology underpinning Ai2's fully open-source byte-level language model Bolmo has been published in the journal Nature.
International journals such as Nature and Cell require authors to declare all image processing steps and prohibit selective enhancement of local features.
The project's reward function and airborne takeoff design reference the Swift paper published in Nature in 2023, which enabled an autonomous drone to defeat human champion racers in real competitive racing.
In 1986, Rumelhart, Hinton, and Williams published the paper "Learning representations by back-propagating errors" in the journal Nature.
Research results related to SynthID Bio were published in the journal Nature.
A related paper published in Nature in 2024 showed that repeatedly fine-tuning a model on its own outputs systematically causes the loss of tail information from the original data distribution.
The position of most mainstream journals such as Nature and Science is that AI tools may be acknowledged in the acknowledgments or methods section, but cannot be listed as authors.
The Strigo AI Atracus introduced in a new Nature paper defeated top human players with a record of 15 wins, 1 loss, and 4 draws across 20 games, through self-play reinforcement learning and test-time search.
12 more timeline events
All Facts (20)
A 2016 survey by Nature showed that more than 70% of researchers had attempted to replicate others' experiments and failed.
90%Verified模型坍缩（Model Collapse）概念在2023年由牛津大学和剑桥大学研究团队在Nature论文中正式提出
70%VerifiedIn early 2023, top journals such as Science and Nature explicitly prohibited listing AI as a paper author, but allowed disclosure of AI's assistive use in the methodology section.
65%UnverifiedNature、Science等顶级期刊已明确要求作者在投稿时声明AI工具的使用情况
90%UnverifiedShumailov等人2024年在Nature上发表的研究证实，即使低至千分之一的合成数据污染也能引发词汇、句法和语义多样性的可测量退化
90%UnverifiedDeepMind published research in Nature in 2021 in which AI discovered the functional relationship between knot hyperbolic volume and algebraic invariants.
60%UnverifiedNature and Science explicitly stipulate that LLMs cannot be listed as paper authors and require authors to disclose their use of AI tools.
60%UnverifiedGoogle DeepMind applied reinforcement learning to chip floorplanning in 2019 and published the findings in Nature in 2021, demonstrating that AI can surpass human experts on specific EDA sub-tasks.
60%UnverifiedTop journals such as Nature have begun requiring authors to submit runnable Notebooks as supplementary materials for papers.
60%UnverifiedGNoME is a project published by Google DeepMind in Nature in 2023 that uses graph neural networks to predict the stability of inorganic crystal structures, producing about 2.2 million new crystal structures at once, of which about 380,000 were assessed as stable candidates.
50%UnverifiedGoogle 于 2021 年在 Nature 上发表了使用强化学习优化芯片布局布线（chip floorplanning）的研究
50%UnverifiedInternational journals such as Nature and Cell require authors to declare all image processing steps and prohibit selective enhancement of local features.
50%UnverifiedThe project's reward function and airborne takeoff design reference the Swift paper published in Nature in 2023, which enabled an autonomous drone to defeat human champion racers in real competitive racing.
50%UnverifiedIn 1986, Rumelhart, Hinton, and Williams published the paper "Learning representations by back-propagating errors" in the journal Nature.
50%UnverifiedResearch results related to SynthID Bio were published in the journal Nature.
50%UnverifiedA related paper published in Nature in 2024 showed that repeatedly fine-tuning a model on its own outputs systematically causes the loss of tail information from the original data distribution.
50%UnverifiedThe position of most mainstream journals such as Nature and Science is that AI tools may be acknowledged in the acknowledgments or methods section, but cannot be listed as authors.
50%UnverifiedThe Strigo AI Atracus introduced in a new Nature paper defeated top human players with a record of 15 wins, 1 loss, and 4 draws across 20 games, through self-play reinforcement learning and test-time search.
50%UnverifiedTop journals such as Nature and Science, as well as most publishers, explicitly require authors to disclose their use of AI tools, and emphasize that AI must not be listed as a paper author.
50%UnverifiedAnna Goldie and Azalia Mirhoseini's most representative work was published in the journal Nature in 2021, with the paper titled "A graph placement methodology for fast chip design."
50%