Linux, finally for everyone

56 points by jjb


theelx

Wow.... this better be satire because I definitely don't like where this is going. I'm usually pro-AI but this is a next level of risk. Giving AI full control of production machines is incredibly dangerous. What happens if the AI hallucinates an issue and does rm -rf to fix it but messes up the variable expansion and deletes the home directory like Steam did (without AI) a while back? That code is certainly in the training data. I doubt even backups are fully safe, and the 99.7% agent uptime doesn't inspire confidence. I see they say that Solaris DevOps uses it; I don't think there's any chance anymore that I want to join that team. Unless there's a new incredibly safe and isolated way to do this that I'm missing, this seems wildly irresponsible to me.