The central argument is that AI automation follows what can be reliably graded, not a simple divide between technical and “human” work. Tasks with clear answers and fast, repeatable checks - such as code that compiles and passes tests, or a voice clone that can be compared with a reference - are especially amenable to machine learning. This explains why AI’s abilities form a jagged frontier: some tasks that seem deeply human are easy to automate, while others remain difficult. The piece maps work across two dimensions: whether it can be graded and whether its value is objectively agreed upon. Coding ranks high on both; poetry is valued but hard to grade; parenting is valuable but not easily scored; and much craftwork resists both measurement and consensus.
Technical feasibility alone does not determine what gets automated: commercial incentives matter. Marketing images, copywriting, and translation have paying clients and briefs that make value easier to agree on, putting creative work used for income at risk even when art itself remains difficult to judge. A survey cited reports that a quarter of illustrators and more than a third of translators had already lost work to generative AI. The piece compares this shift with power tools displacing traditional carpentry, arguing that workers may be pushed from making things toward operating machines. Its proposed distinction for adapting is that AI produces deliverables - code, reports, images - while people achieve outcomes through judgment, accountability, and social coordination.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.