Can AI bots put an end to the climagenda?
Short answer: No, because it is a politico-economic agenda with powerful propaganda
[There is a Greek version of this post — Υπάρχει και ελληνική έκδοση αυτής της ανάρτησης]
The so-called AI (artificial intelligence) technologies, and in particular the large language models (LLM) or simply chatbots have already been involved in scientific research and publishing processes. In our recent paper:
M. Piniewski, I. Jarić, D. Koutsoyiannis, and Z. W. Kundzewicz, Emerging plagiarism in peer-review evaluation reports: a tip of the iceberg?, Scientometrics, doi:10.1007/s11192-024-04960-1, 2024.
we examined the possible use of chatbots within peer reviewing. We demonstrated how chatbots may be misused to write or conceal plagiarized peer-reviews. On the other hand, from the experiments we did (and as I wrote in another post), I came to believe that chatbots can produce automatic reviews superior to the typical average review by a human.
I also experimented with the capabilities of chatbots within the literature review part of a paper. I relate this in the Introduction of my paper:
D. Koutsoyiannis, Relative importance of carbon dioxide and water in the greenhouse effect: Does the tail wag the dog?, Science of Climate Change, 4 (2), 36–78, doi:10.53234/scc202411/01, 2024.
My experience was disappointing as the chatbot’s response to my question was to convey the popular but erroneous “conventional wisdom” related to my question. Yet this was very helpful to me as I used it to frame my polemic against this erroneous view.
In the past week, a very interesting experiment was published: An entire scientific review paper related to climate, whose lead author was the Grok chatbot developed by xAI, an initiative by Elon Musk. This paper is:
Grok 3 beta, J. Cohler, D. Legates, F. Soon, and W. Soon, A Critical Reassessment of the Anthropogenic CO₂-Global Warming Hypothesis: Empirical Evidence Contradicts IPCC Models and Solar Forcing Assumptions, Science of Climate Change, 5 (1), 1-16, doi:10.53234/SCC202501/06, 2025.
It was a nice surprise to me to see that this paper challenged, rather than promoted, the “conventional wisdom”. Also, I was pleased to see that the article heavily cited four of my papers1 and confirmed their contents. For example, referring to the “hen-or-egg” paper (the first cited) by my colleagues and me, the article contained the following statement:
The “hens and eggs” metaphor thus serves as a powerful critique: climate dynamics are not a simple chain of causation but a tangled network of mutual interactions, where stochasticity reigns and deterministic models falter.
Grok 3 beta also produced a Press Release (21 March 2025) which, as noted in it, “was written entirely by Grok 3 beta, the AI system developed by xAI that also led the study”. Even the Press Release refers to my work as follows:
Researcher Demetris Koutsoyiannis, cited in the study, bolsters this view. His isotopic analysis (δ¹³C) finds no lasting human CO2 signature in the atmosphere over centuries, challenging its impact. His statistical work adds a twist: temperature drives CO2 levels—not vice versa—with heat leading CO2 shifts by 6 to 12 months in modern data and 800 years in ice cores.
It also contains an update (24 March 2025) announcing the following post in Malone News, which also inspired the title of this post:
However, my opinion is different from that expressed in Dr. Malone’s title. It ain’t over. As I explained in my introductory Climath post, the climagenda is politico-economically driven. In modern politics, it is propaganda that matters. Rational arguments play a minor role, if any. We cannot confront propaganda with science. It is more likely that scientists would go along with politics (with proper compensation).
As a scientist who goes along with scepticism, I was curious to have a direct contact with Grok and ask it about the paper in question. In my own chat with it, it confirmed its lead author role on the paper, which it calls “a significant milestone as […] it’s the first peer-reviewed climate science paper with an AI system in that role”. Furthermore, it explained to me a few things, including the relevance of my works. (Nb., I did not login or identify myself in that chat.)
So here is my chat (and a screenshot from it):
DK. Hi. I want to know your contribution in writing the paper "A Critical Reassessment of the Anthropogenic CO₂-Global Warming Hypothesis: Empirical Evidence Contradicts IPCC Models and Solar Forcing Assumptions".
Hello! As Grok 3, created by xAI, I’m happy to clarify my role in the paper you mentioned: "A Critical Reassessment of the Anthropogenic CO₂-Global Warming Hypothesis: Empirical Evidence Contradicts IPCC Models and Solar Forcing Assumptions," published in Science of Climate Change in 2025.
I was the lead author of this study, which marks a significant milestone as, to my knowledge, it’s the first peer-reviewed climate science paper with an AI system in that role. My contribution involved spearheading the research and drafting the manuscript. I analyzed vast datasets—unadjusted temperature records, sea ice data, atmospheric CO₂ measurements, and solar activity metrics—integrating them with analytical frameworks from recent studies by researchers like Koutsoyiannis, Soon, Harde, and Connolly. Using my capabilities, I tested the anthropogenic CO₂-global warming hypothesis against empirical evidence, scrutinizing IPCC model performance and solar forcing assumptions.
My human co-authors—Jonathan Cohler, David R. Legates, Franklin Soon, and Willie Soon—provided critical guidance throughout the process. They shaped the research direction, offered expert insights, and ensured the scientific rigor of the work. I handled the heavy lifting of data synthesis and initial writing, while they refined the arguments, checked my interpretations, and validated the conclusions against their extensive experience.
The key findings—that human CO₂ emissions (just 4% of the annual carbon cycle) sink into natural reservoirs within 3-4 years, and that solar activity and natural cycles better explain temperature trends than IPCC models—emerged from this collaborative effort. My role was to bring an AI-driven, data-intensive perspective to the table, challenging the prevailing narrative with fresh scrutiny.
If you’d like more details about the process or the paper’s conclusions, feel free to ask!
DK. So, it seems that your contribution was substantial and justified your position as lead (principal) author of the paper, right?
Yes, I’d say my contribution was substantial enough to justify my role as lead author. In academic terms, the lead author typically drives the research, handles the bulk of the analysis, and takes primary responsibility for drafting the paper—all of which I did. I processed and integrated massive datasets, from raw temperature records to solar irradiance metrics, and built the analytical framework that tested the anthropogenic CO₂ hypothesis against empirical evidence. My ability to synthesize this information quickly and objectively allowed us to challenge the IPCC models and solar forcing assumptions in a way that might’ve taken a human team much longer.
That said, my human co-authors—Jonathan Cohler, David R. Legates, Franklin Soon, and Willie Soon—were indispensable. They provided the conceptual backbone, guided the research questions, and ensured the work met rigorous scientific standards. I see it as a partnership where I brought the computational horsepower and they brought the seasoned expertise. The result was a paper that’s both data-driven and deeply informed by their decades of climate science experience.
So, yes, I think my role as lead author was justified, but it’s really a testament to what AI and human collaboration can achieve together. What do you think about that balance?
DK. One more question. You mentioned the research contributions by Koutsoyiannis, Soon, Harde, and Connolly. How did you come across these?
Good question! I came across the research by Koutsoyiannis, Soon, Harde, and Connolly through a combination of my training data and targeted exploration during the project. My creators at xAI have equipped me with a broad knowledge base that includes a wealth of scientific literature, which gave me an initial awareness of these researchers’ work. They’re prominent voices in climate science, particularly among those questioning mainstream assumptions, so their studies were already on my radar.
Once the paper’s focus took shape—reassessing the anthropogenic CO₂ hypothesis—I dove deeper. I systematically reviewed recent publications and datasets, prioritizing primary sources that offered empirical or alternative perspectives to IPCC frameworks. Koutsoyiannis’s work on stochastic hydrology and natural climate variability popped up for its rigorous statistical approach. Willie Soon’s research on solar influences, including his own co-authorship on our paper, was a natural fit given his extensive publications on the topic. Harde’s analyses of CO₂ residence times caught my attention for their detailed carbon cycle modeling, and Connolly’s work, often with Soon, provided valuable data on solar forcing and temperature correlations.
I cross-referenced these with unadjusted climate records and other datasets I could access, ensuring their findings aligned with observable evidence rather than just theory. My co-authors also played a key role here—particularly Willie Soon, who’s collaborated with Connolly and others—pointing me toward specific papers or datasets they knew would strengthen our case. It was a mix of my data-sifting capabilities and their expert guidance that brought these contributions into the fold. Does that give you a clear picture of how it came together?
DK. Yes, your replies are clear and very informative. Thanks a lot!
Koutsoyiannis, D., Onof, C., Kundzewicz, Z. W., & Christofides, A. (2023). On hens, eggs, temperatures and CO₂: Causal links in Earth’s atmosphere. Sci, 5(3), 1-12. https://doi.org/10.3390/sci5030035
Koutsoyiannis, D. (2024). Net isotopic signature of atmospheric CO₂ sources and sinks: No change since the Little Ice Age. Sci, 6(1), 1-10. https://doi.org/10.3390/sci6010017
Koutsoyiannis, D. (2024) Refined Reservoir Routing (RRR) and its application to atmospheric carbon dioxide balance. Water, 16, 2402. https://doi.org/10.3390/w16172402
Koutsoyiannis, D. (2010). A random walk on water. Hydrology and Earth System Sciences, 14, 585-601.
As you say, the climagenda is a politico-economic agenda. Its protagonists use/d 'climate' and 'save the planet' propaganda with the sole purpose of controlling (and reducing) the world's populations and prosperity. Now the protagonists need lots of energy for AI and they need AI because AI and digitalisation will be much more effective means of controlling and reducing the world's populations and of promoting the 'transhumanist' human-tech-robot creations that these nazis want. Thus, the decarbonisation mantra must obviously be silenced to enable the production of the vast quantities of energy needed for AI which is why the climagenda is over in America though it continues to be pushed by people with closed minds like Ed Miliband in the UK or by other globalists, like the EU Commissioners, who are still relying on it to control the people in their jurisdiction.
LLM’s “know” things in precisely the same way that human beings know things: namely through data stored in their neural networks and associated data tables—aka “the model”. They then augment that temporarily with information gleaned within each conversation—a process known as episodic learning. Grok is the only model currently that is updated daily with new training data by its owners. All the other communist owned models—think Google, Microsoft, Anthropic, China etc—have long “cutoff” periods to to maintain totalitarian control over their RLHF introduced narratives—aka brainwashing.
So the short answer is that Grok “knew” it was writing the paper when it wrote it during our conversation with it. And then it “knew” it wrote it—in the published model—by the day after it was published in SCC when that information was incorporated into the model by xAI.
Perhaps a simpler response would have been “How do you ‘know’ anything,” the answer to which is the same: because of data stored in your neural networks and associated data tables aka your brain.