> Nowadays, you can specify the task you want to automate, fire off Codex or Claude Code, check back in 30 mins and have an almost-perfect result.
It’s clear the author has not been using LLM’s enough
Spent a few hours the other day trying to get Claude to produce some spring-boot code. Match the library versions to behavior and between each other. The thing that it should be perfect for. Nope. The moment I’m not explicit, it doesn’t even put version numbers into POM. Until it randomly does. Easier and faster to take the time, read the docs and do it myself.
Talk about a low effort post. This is LinkedIn worthy.
I was thinking of a nice way to phrase this
And so continues our plunge towards losing ownership of the means of computation
That only means that LLMs suffer from a discoverability issue. An average user would need to know if it's worth paying the monthly cost for their circumstances, not pay and then find ways of justifying it.