I’ve been noticing more tools that turn a half-finished thing into something that looks ready with almost no extra effort. That sounds great until you realize the last bit of judgment is where a lot of mistakes get caught.
Has anyone found a good line between helpful automation and the point where it starts skipping the part of the workflow that still needs a human?
“Done” starts feeling fake the second the tool makes something look shippable before I’ve done the “wait, is this actually correct? ” pass. My line is basically autocomplete vs autopilot. I’m happy to let AI fill the boring gaps, but I still force a last pass that’s purely hunting for what’s wrong (more like a QA sweep than a polish sweep). The moment I skip that judgment pass because the output looks finished, that’s when the dumb mistakes sneak in and I get that uneasy “this was too easy” vibe.
:: Copyright KIRUPA 2024 //--