Because the code was never the hard part?

it was kind of a hard part, but not the hardest

Because of the 1% problem: https://try.works/coding-agents-and-the-growing-1-problem

humans are still very needed to verify AI outputs

indeed, why _hasn't_ productivity? edge of my seat

Because it writes wrong code 100x faster too and humans trying to make sense of it all

it’s not always wrong. some of it is wrong. the trick is figuring out what’s actually correct

One angle that doesn't come up enough in this discussion: even when productivity gains don't materialise, engineers are still feeling the pressure as if they have. In our burnout survey data, AI pressure to do more has emerged as a top 4 burnout driver in 2026 — a finding that didn't exist two years ago. So the productivity paradox has a human cost attached to it. CEOs see no productivity gains, but engineers are burning out from the expectation that they should be more productive because of AI. The pressure exists regardless of whether the output does. We wrote this up here if anyone's interested: rechargedaily.co/blog/ai-productivity-burnout-paradox Also running an anonymous survey on this — 3 mins, results public: https://docs.google.com/forms/d/e/1FAIpQLSdu-1Sa6oPvhDtFtBuK...