AI job vulnerability depends on workflow stages, not just output categories

georgemillo · x · 2026-09-01

Eliot Ranch argues that assessing job vulnerability to AI based solely on output similarity is flawed. The real risk lies in specific workflow stages. To use AI effectively, one needs both: 1) expertise in the task, and 2) skill in prompting, context design, and setting guardrails. This dual requirement explains why AI thrives in coding, where the predictable structure and developer mindset align perfectly with this workflow.

Original post →

More from AGI Musings

AGI Musings channel →