Post

I don't think the job has fundamentally changed, really, judgement about what's worth doing has always been the most important thing for most work.

Also to repeat another critic: No one has had to maintain this stuff yet, or seen what the drawbacks of your engineers not really understanding the codebase are. The knowledge-rot may prove organizationally fatal.

I didn't say the change would be good! I do think it's a tool that will empower good engineers to be even better, and conversely people who weren't careful before are going to become that much more dangerous going forward, for exactly the reasons you just said

I'm also not even convinced that these are yet "indispensable" and that engineers must adapt-or-die here. I see a lot of excitement but the velocity increase does not seem like it's sustainable, especially when you factor in inference costs.