Conversation

Notices

1. Embed this notice
   AndresFreundTec (andresfreundtec@mastodon.social)'s status on Friday, 09-Jan-2026 05:16:17 JST AndresFreundTec

   Hoped for effect of AI use of new postgres contributors: Faster rampup, higher quality patches.

   Real effect: A torrent of pointless patches, which often don't even pass tests. A torrent of AI generated pointless review comments.

   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Friday, 09-Jan-2026 20:52:29 JST 翠星石

     in reply to

     • tyil
     @tyil @AndresFreundTec If you actually did any research into the topic, you'll realize that all LLM's is copy combinations of terrible code that is full of bugs and security vulnerabilities and therefore all that results is development majorly slowed down, as cleaning up terrible code is slower than just writing non-terrible code (of course, copying good code normally is faster and more reliable).
   • Embed this notice
     tyil (tyil@fedi.tyil.nl)'s status on Friday, 09-Jan-2026 20:52:32 JST tyil

     in reply to

     @AndresFreundTec@mastodon.social Hoped for effect of AI use of new postgres contributors: Faster rampup, higher quality patches.How could you have been hoping for that? If you know how "AI" (more accurately, generative LLMs) work, you'd know they'd never be able to make "higher quality patches".
     
     Even if you don't know how these systems work, and have no desire to understand, you could look at any place where generative LLMs are used, and see that they do not produce anything worthwhile, so it would seem silly to me to believe they'll produce anything worthwhile in a particular project out of sheer hopr.