This morning's Hacker News front page led with a Vatican encyclical. *Magnifica Humanitas* — 1510 points, 846 comments — Leo XIV's first major teaching document, on AI. A few slots down, a senior engineer arguing he writes better code more slowly with agents. 847 points. Same day, Uber's president told investors AI spending is "getting harder to justify."
Three signals, three rooms. The AI conversation has quietly turned a corner — from capability to value, from "what can it do" to "is it worth what it costs."
For three years the argument was capability. New model, new ceiling, new "you have to see this." The skeptic was assumed to be uninformed. The corpus today is doing something different. Nolan Lawson — who uses agents daily — argues the productivity story is upside down: agents force articulation, slow down shortcuts, produce better code at lower throughput. The encyclical takes another tack: it argues from a theology of the human person and lets the regulatory ask follow. Every AI debate is anthropological in disguise; this one is honest about it. Uber is pricing the same question in finance terms. None of these are anti-AI. That's the point. They're three independent attempts to ask what we're optimizing for.
The obvious objection is that this is what always happens when a hype cycle exhausts itself — capital tightens, the believers regroup, the conversation matures. True. But the turn has a sharper edge this time. The "for what?" question isn't being asked by skeptics on the sidelines. It's being asked by users at the workbench, by buyers writing checks, and now, formally, by an institution with two millennia of practice on the human question. Concession granted: this could be the bottom of a cycle. The relevant claim is that the bottom is where the real arguments start.
A plumb line measures true vertical. For three years the AI conversation has been pulled visibly off-axis by everyone with something to sell. The line is finally being held still long enough to read.