AI finished the task, so why isn't it what I wanted?
This is a commentary published on the WeChat account “我是张大朵啊” by Zhang Qiduo, prompted by another essay, “Thoughts on AI-Native Organizations.” The author argues that when AI takes over more work, the output often gets “done” without being what the user actually wanted, and attributes the difference to accumulated personal context — she says her years of dialogue with her own AI, which she describes as more than ten million words, taught it her standards of judgment, while a company's internal AI lacks that history. The piece also argues that AI-native organizations push employees to deposit decisions, discussions and judgments as organizational assets, which can standardize individual skills and reduce an organization's dependence on any one person, and suggests individuals build their own production systems.
It reframes “context is value” as a question of labor and distribution: who owns the experience an individual accumulates by working with AI.