The machines are fine. I'm worried about us (ergosphere.blog) AI

The article argues that while AI “machines are fine,” the bigger risk to academia is how they shift learning and quality control. Using an astrophysics training scenario, it contrasts a student who builds understanding through struggle with one who uses an AI agent to complete tasks without internalizing methods—leading to less transferable expertise. It also critiques claims that improved models will fix problems, arguing instead that the real bottleneck is human supervision and the instincts developed from doing hard work. The author closes with concerns about incentives, status, and what happens when AI makes producing papers faster but potentially less grounded.