The standard story is that AI frees PMs from busywork so they can focus on strategy. That is comforting and half true. The harder truth sits underneath it: a large share of the busywork was never busywork to the person doing it. It was how they proved they deserved the next level.
The mechanism
Think about how a PM actually gets promoted. Rarely on "she made the number go up," because attribution is contested and the number moved for ten reasons. Usually on "she drove the strategy for X," and the proof of driving is a document with her name on it. The artifact is legible. Impact is arguable. Calibration committees run on the legible thing, so for years the quiet incentive was to produce impressive artifacts. Not fraudulently. The system taught PMs that the doc was the job.
AI removes the artifact. When a spec collapses from three weeks to an afternoon, you do not just save time. You remove the object the ladder was calibrated to measure. The prototype and the eval are a better artifact for the customer and a worse one for the promotion system, because they are small and fast and do not carry the visible weight a committee learned to read as seniority. The evidence got better and less legible at the same moment, and legibility is what the ladder was built on.
Why it is painful, not just inconvenient
The replacement evidence, outcomes and judgment, is harder to see and harder to game. That is a feature and the reason most orgs will avoid rebuilding their leveling. It is easier to keep promoting the best document producers into a role where documents no longer exist. The correction gets deferred until a reorg forces it. If you are a PM, the move is to stop treating the doc as your evidence and start building the kind the new ladder will read: shipped surfaces with metrics, evals you own, decisions with reasoning attached. Then ask your manager what gets you to the next level when the deliverable is gone. Their answer, or lack of one, tells you where you actually stand.