Why Closed Models Are Bad for Scientific Research

shangbinfeng · x · 2026-07-16

This repost emphasizes that **closed models are detrimental to scientific research**. The core argument is that researchers cannot know the internal structure or update details of models that are constantly changing. Additionally, every evaluation incurs direct financial costs, making experiment reproduction and continuous comparison much harder. The post notes that this is a co-authored article on the issues with closed models, pointing out that this problem will only become more critical over time.

Original post →

More from Research

Research channel →