i stumbled onto this thought abt why we cant just rely on manual audits anymore. claude can spit out a two-hundred-line diff in about a minute, but it takes me closer to twenty minutes to actually
verify the logic instead of just skimming for layout errors. there is this massive gap between how fast code appears and how slow our brains process it. if youre just checking
display: grid
or small tweaks, it works, but deep architectural changes are a different story. ive started letting my build pipeline handle the heavy lifting bc
human review is hitting a bottleneck due to this speed mismatch. we used to pride ourselves on reading every line, but now were just
scrutinizing checking for structural shape.
it's basically impossible to maintain high standards when the volume of plausible-looking junk scales this fast. does anyone else feel like theyre becoming a glorified linter instead of an actual engineer?
https://dev.to/indiecoredev/i-dont-review-ai-code-my-build-does-436d