Submission (Preprint #9446) that references a study performed but provides no results - accept or reject?
Daniel Segessenman
Public
Seen by 75
Hello Everyone,
I was not sure with how to proceed on this submission currently in the repository manager: Preprint #9446 - Leveraging Machine Learning for Predictive Modeling of Geohazards: A Data-Driven Approach in Earth Sciences
It references a 'study' that was apparently completed by the author, but has no results or figures. It goes straight from a methods/materials sections to conclusions that have statements like: "This study has demonstrated that data-driven models, when properly implemented, offer significant advantages over traditional techniques in terms of accuracy, adaptability, and speed." There are no data, tables, figures, described analyses, or links to any data that demonstrate that there was any study performed. It seems more like a grant proposal project description, but written in the tense of a completed study. It also generally lacks in-text citations (there is one) and a cover-sheet (which I will add if we think this should be approved). I'm leaning toward reject, but it was unusual enough that I wanted to get other opinions.
Cheers!
Colleen Rosales Wed 18 Jun 2025 6:22PM
That sounds like a reject to me...
Daniel Segessenman Wed 18 Jun 2025 9:45PM
Roger that, thank you for the input! I will reject and suggest Zenodo to the author.
Bruce Caron Wed 18 Jun 2025 11:27PM
Claude 4 gave me this: Looking at this academic paper, there are several indicators that suggest it was likely written by AI:
Strong indicators of AI authorship:
Generic, formulaic structure - The paper follows a very standard academic template without the natural variations you'd expect from human writing
Lack of original research - Despite claiming to be a study with methodology and results, there are no actual experimental data, specific results, or novel findings presented
Vague, sweeping claims - Statements like "Results indicate that ML-driven models significantly enhance prediction accuracy" without any supporting data or specifics
Repetitive language patterns - Overuse of phrases like "significant," "comprehensive," "robust," and similar academic buzzwords throughout
Missing crucial elements - No actual results section, no specific numerical findings, no detailed methodology despite claiming to have one
Surface-level treatment - The content reads like a synthesis of existing knowledge rather than original research, with no deep insights or novel contributions
Additional concerns:
The references appear to be real papers, but the way they're integrated suggests summarization rather than genuine scholarly engagement
The writing is technically correct but lacks the authentic voice and specific expertise you'd expect from a geosciences researcher
The methodology section describes what would be done rather than what was actually done
While I can't be 100% certain without additional analysis, the combination of these factors strongly suggests this was generated by AI rather than written by a human researcher conducting original work.
Bruce Caron Wed 18 Jun 2025 11:31PM
@Daniel Segessenman Great job on this one!
Bruce Caron · Wed 18 Jun 2025 6:18PM
You are right... this is the description of how the study was done. It lacks the actual research results. I would suggest to the author that Zenodo might be a better spot for this.