1367% Increase in AI Climate Change Paper Abstracts: 2018 to 2025

As one of the most pressing global challenges, climate science relies on clear, accurate, and credible communication of research findings, often in the form of formal scientific articles.

So, what happens when the integrity of those papers’ abstracts is called into question?

The emergence of large language models (LLMs) like ChatGPT has made it increasingly easy for researchers to draft scientific content, including abstracts, with minimal effort. However, generative AI tools aren’t perfect and have been known to hallucinate and make mistakes.

Without proper oversight, it’s therefore possible that AI-generated climate change abstracts may not be as accurate as they should — or need — to be. 

Since it’s often the first (and in some cases, only) portion of a paper read by policymakers, media, and the general public, maintaining the integrity of abstracts is crucial not only academically, but also socially and politically.

Due to this influence, this study examines the prevalence of AI-generated abstracts in climate change-related publications from 2018 to 2025 and assesses implications for the growing role of AI in scientific authorship.

Objectives of the Study

Based on thousands of climate change abstracts collected from 2018 to 2025, this study aimed to:

• Find out how common AI-generated climate change abstracts are in 2025
• See how that rate has changed since 2018
• Explore what this trend means for the role of AI in academia, the potential need for transparency in AI-assisted writing, and the ethical implications for scientific authorship

Key Takeaways (TL;DR)

• AI climate change abstracts in related scientific publications increased by 1367% from 2018 to 2025
• The rate remained under 6% from 2018 to 2020, hit 16% in 2021, then stayed relatively stable until rapidly increasing to 43.16% in 2024
• In 2025, 45.83% of climate change abstracts are likely AI-generated
• Such a dramatic spike suggests an increasing reliance on AI tools to generate abstracts in scientific writing

1367% Increase in AI Climate Change Abstracts: 2018 to 2025

We found that the proportion of AI climate change abstracts in scientific journals rose dramatically from 2018 to 2025. 

Check out our quick overview chart to see the rates of Likely AI climate change abstracts by year.

Then, keep reading for a year-by-year discussion of the data.

‍

The rate of AI-generated climate abstracts stayed fairly low and steady from 2018 to 2020:

• In 2018, only 3.13% of abstracts were identified as likely AI-generated
• This increased to 5.49% in 2019
• Then, the rate held relatively steady in 2020 with a moderate drop to 5.10%

Then, overall, despite a slight drop in 2022, the trend began to accelerate:

• In 2021, the rate tripled to 16.00%
• Although there was a slight decline in 2022 to 13.00%, the rate continued to rise in the following years
• In 2023, 17.20% of abstracts were flagged as AI-generated
• This was followed by a dramatic spike to 43.16% in 2024
• By 2025, the rate reached an all-time high of 45.83%

This means that by 2025, nearly half of all climate change abstracts now contain content that is likely generated with the assistance of AI.

Overall, the increase from 2018 to 2025 amounts to a 1367% rise in the rate of AI-generated abstracts. 

This exponential growth underscores the rapid adoption of AI tools in academic writing, particularly following the release of ChatGPT at the end of 2022.

In 2025, 45.83% of Climate Change Abstracts Are Likely AI

Although the sharp rise in AI-generated abstracts from 2018 to 2025 reflects a fundamental shift in how scientific knowledge is being produced and communicated, this trend isn’t limited to academia.

The use of generative AI tools has increased across many industries during this period, especially in online reviews. Previous studies we have conducted found growing rates of likely AI-generated bank reviews, Glassdoor reviews, and lawyer’s office reviews.

So, the finding that 45.83% of climate change journal abstracts in 2025 are likely AI-generated highlights a broader pattern. 

It also raises important questions about authorship, accountability, and academic publishing standards.

Implications of AI Content in Climate Change Abstracts

The prevalence of AI climate change abstracts has several significant implications:

• Authorship, transparency, and attribution: If AI is incorporated, its use should be transparently disclosed (Read more about how to cite AI as a source.)
• Research integrity and fact checking: Steps should be taken to properly fact-check content to avoid AI hallucinations or inaccuracies in abstracts 
• Peer review and editorial standards: Journals or reviewers should take steps to identify and evaluate likely AI content, such as through AI detection tools.

Ultimately, researchers, journals, and institutions will likely need to reconsider policies around AI disclosure, authorship standards, and peer-review processes to preserve the credibility of climate change research and other scientific disciplines.

Final Thoughts

This study reveals an exponential rise in AI-generated climate change abstracts from 2018 to 2025. Considering the potential implications of this trend, such a significant increase underscores the need for academics, researchers, and publishers to manage the ethical and practical implications of AI in academic writing. 

To uphold transparency, trust, and integrity in scientific communication, those in the academic community should consider:

• Disclosing AI use clearly and consistently when generative tools are used to create abstracts or other parts of a paper
• Establishing clear authorship guidelines that define the role and limits of AI use in scientific publishing
• Using AI detectors on submitted articles to flag likely AI content for further review before publication
• Develop and update peer review standards to reflect and address the risks of AI hallucinations and errors

Until more robust guidelines are developed regarding generative AI use in climate change and other scientific disciplines, readers may also consider evaluating abstracts more critically themselves. 

Do you think you’re reading an AI-generated abstract? Try Originality.ai’s AI Checker to find out today.

Curious about the trend of AI-generated content on other platforms? Read more:

• Human-Written Websites are Worth 39% More Than AI Websites
• 1361% Increase in AI Temu Reviews From 2022 to 2025
• 94.9% of US Bank Reviews Are Human-Written in 2024
• 4600% Surge in AI Canadian Online Bank Reviews: 2018 to 2025

Methodology

To study the prevalence of AI-generated abstracts in climate change research, we analyzed scientific abstracts published from 2018 to 2025. Abstracts were collected using the OpenAlex API, filtered by the keyword “climate change,” publication year, and presence of non-empty abstracts, with up to 500 entries per year. A custom function reconstructed the abstracts from OpenAlex's inverted index, preserving key metadata.

Each abstract (minimum 50 words) was evaluated using the Originality.ai API, which provided an AI-likelihood score and binary classification. Scans were repeated if API errors occurred.

All data — including titles, abstracts, metadata, and AI detection results — were compiled into a CSV file using pandas. This dataset supported further analysis, including trends in AI-generated content over time.