Comment by miningape
1 day ago
Don't really have anything to add but I do want to say you're not alone - I feel very similarly about AI tooling, the level of satisfaction I get from using them (none), the need for interesting technical challenges, etc. etc.
There are dozens of us!
Re: AI, that's not to say I don't use it, I just view it as a sometimes useful tool that you have to watch very closely. I also often view their use as an X-Y problem.
Another recent example: during the same AI week, someone made an AI Skill (I'm not sure how that counts as software, but I digress) that connects to Buildkite to find failed builds, then matches the symptoms back to commit[s]. In their demo, they showed it successfully doing so for something that "took them hours to solve the day before." The issue was having deployed code before its sibling schema migration.
While I was initially baffled at how they missed the logs that very clearly said "<table_name> not found," after having Claude go do something similar for me later, I realized it's at least partially because our logs are just spamming bullshit constantly. 5000-10000 lines isn't uncommon. Maybe if you weren't mislabeling what are clearly DEBUG messages as INFO, and if you didn't have so many abstractions and libraries that the stack traces are hundreds of lines deep, you wouldn't need an LLM to find the needle in the haystack for you.