Artificial intelligence systems were better than human scientists at spotting technical flaws in clinical research papers, according to a new study. The finding points to a possible role for AI in checking the quality of studies whose results can shape medical decisions.
Clinical studies matter because doctors, regulators and patients rely on their conclusions when choosing treatments. Errors in how a study is designed or how its data are analyzed can undermine those conclusions, which is why careful review before and after publication is so important. The study found that the AI systems outperformed the human scientists in identifying these kinds of errors in the papers it examined.
The results leave several questions open. The research does not make clear how far the findings extend beyond the specific papers that were tested, so it is uncertain whether the same advantage would hold across other fields or types of research. It is also unresolved whether AI should serve as an assistant to human reviewers or eventually take over parts of their work.
For now, the study suggests that AI tools could become a useful aid for scientists and journals trying to catch problems in medical research, while leaving open how much responsibility they should be given in the review process.
Credit. Based on reporting for Science Magazine News. Is this your work? Write to us.
34
