Following a post dated January 4th of the current year, accessible at the aforementioned link, where an interesting and significant study by researchers from the renowned UCLondon was disclosed, showcasing the remarkable ability of the Generative Artificial Intelligence model, GPT-4, to produce scientific articles in less than 60 minutes, I take the opportunity to share a very recent study. In this study, North American researchers analyzed the capability of the same GPT-4 to generate a review article in the field of biomedicine https://www.biorxiv.org/content/biorxiv/early/2024/04/17/2024.04.13.589376.full.pdf
Regarding the implications of this new reality, where the production of scientific articles will no longer be exclusive to researchers, leading to an evident decrease in the value of those articles in evaluating researchers' performance, it is worth revisiting the previous post "AI has radically changed the core university business, shifting focus from teaching and publications to “assessment, curation, and mentoring”
In this context, it is important to note that just yesterday, an article featured in the latest edition of The Economist highlighted a significant increase in the capabilities of Generative AI models. This suggests that these models will likely have substantially enhanced abilities in the future when it comes to generating scientific articles. https://www.economist.com/science-and-technology/2024/04/17/large-language-models-are-getting-bigger-and-better
PS - The senior scientist in the aforementioned study is a Computational Biomedicine professor named Jason H. Moore (Scopus h-index=82), who not long ago secured funding of $8 million to study new artificial intelligence methods and how they can be applied to find new biomarkers of Alzheimer's disease.
