How citations ruined science (davidoks.blog)

A recent examination by Nature highlights a troubling trend in the scientific community: the surge of low-quality, AI-generated research papers in the biomedical field, with about 15% of abstracts in 2024 likely written by AI. This phenomenon is exacerbated by the overwhelming influx of submissions to journals, driven in part by AI's ability to produce quick content. As peer reviewers also turn to AI tools to manage the unprecedented volume of manuscripts, a vicious cycle emerges wherein shoddy research gets published and subsequently cited, leading to an increasing dilution of scientific integrity. This situation underscores a broader issue within the scientific landscape—namely, the "citation revolution" that has elevated metrics above genuine inquiry and knowledge. Scientists are incentivized to prioritize quantity over quality due to a system that rewards publication rates and citation counts. As a result, AI technologies, which could enhance scientific work, are instead being leveraged to churn out low-quality outputs that masquerade as rigorous research. The AI/ML community must grapple with the implications of this trend and rethink the institutional structures that now govern scientific inquiry, seeking a reconciliation between AI's potential for innovation and the integrity of scientific progress.