ChatGPT genera datos falsos de ensayos clínicos que parecen reales
Italian researchers demonstrated the fragility of scientific integrity by using ChatGPT to generate false yet convincing datasets that support conclusions contrary to existing medical evidence. This experiment reveals that artificial intelligence can create non-existent clinical trial measurements and results in a matter of minutes, posing a serious threat to the validity of academic publications. The relevance to the open data field lies in the urgent need to develop robust verification and transparency mechanisms. Although synthetic data may appear authentic at first glance, in-depth statistical analyses can detect subtle anomalies, such as unnatural age distributions or inconsistent correlations. This underscores that the accessibility of open data should not exempt journals and scientific communities from implementing rigorous quality controls to distinguish between genuine evidence and that fabricated by algorithms. In conclusion, the study warns that this technology facilitates the large-scale production of misleading information, complicating the work of traditional peer review. An immediate update of evaluation protocols and the development of new detection tools are required. Transparency regarding the origin of data thus becomes a fundamental pillar for maintaining the credibility of science in the era of automated content generation.
Source: lavozdegalicia.esPublished on 2023-12-08
Related news
- Las peticiones semanales de desempleo en EEUU suben menos de lo esperado
- Team Abu Dhabi Register Thrilling Last-Ball Win Over Delhi Bulls
- Study: Pay, experience positively impact wildland firefighter retention; competing wages not so much
- Falta de nombramiento de comisionados del INAI debilita transparencia del gobierno | Entérate
- 90 per cent drop in visitor arrivals to Israel in November - IndiaPost NewsPaper