把工作交给 AI 后，为何反而更累？Molly Graham 重写职场积木法则

Delegating execution to an AI agent does not transfer accountability when the human still carries the burden of oversight and final approval.

An engineer can become more productive while grieving the loss of craftsmanship, because steering agents feels different from doing the work directly.

Rising burnout and strong enthusiasm can coexist in a survey, so autonomy and perceived amplification deserve attention alongside aggregate productivity claims.

A team's net gain depends on useful outcomes after review and rework, rather than on the volume of generated drafts or consumed tokens.

Treating an agent as a junior collaborator encourages explicit context, clear acceptance criteria, and accountable review instead of blind trust in fluent output.

Humans should retain judgment, trusted relationships, and strategic intent, especially when they cannot reliably evaluate what a good result would look like.

AI can lower the barriers between professional roles, but removing inherited handoffs does not eliminate legitimate safety, compliance, or production constraints.

A practical learning habit is to pause before a task and ask whether AI can help, without assuming that earlier adopters hold an insurmountable advantage.

Greater ambition is valuable when it produces enduring work that deserves people's time, not merely when cheaper generation makes more products possible.

During rapid change, managers should model accountable AI use, acknowledge emotional strain, and establish standards that protect both people and the quality of work.