If AI can draft anything instantly, maybe the real shortage is people willing to own the decision and take the blame.

This explores whether, as AI makes routine thinking cheap, the scarce and valuable thing in the labor market becomes people who can make consequential decisions, check AI's work, and answer for the outcome. The corpus largely agrees, though with conditions attached. The clearest statement is that when first-pass cognition is cheap but fallible, human work survives where people exercise judgment, verify outputs, and accept accountability. Whether that turns into good jobs depends less on how capable AI is and more on how institutions are designed, especially whether they keep the right to question decisions and the chance to learn by doing What makes accountable judgment scarce when AI cognition is cheap?. An economic framework points the same way. Cognitive tools stand in for the skill of carrying out a task but make two kinds of judgment more valuable: spotting which opportunities to pursue, and judging what an outcome is worth. Only the second consistently gets more valuable as the tools improve. That helps explain why computers widened inequality while AI often narrows it within firms How do cognitive tools reshape the three types of worker skill?.

A related idea treats this as a matter of throughput. The cost of automating work is falling much faster than the cost of checking it, so the human capacity to audit AI output becomes the bottleneck on growth. AI output that nobody has checked becomes a hidden cost that the rest of the economy absorbs What becomes scarce when AI execution grows cheap?. Pushed to the extreme, you get what one note calls 'epistemic hyperinflation': AI produces claims faster than people can evaluate them, confidence in knowledge in general drops, and the tools used to evaluate are themselves AI-generated, so the gap keeps feeding itself Can AI generate knowledge faster than humans can evaluate it?.

Here is the twist: scarce does not mean used. Managers with limited attention rationally hand decisions to machines even when their own judgment would do better, because judgment takes effort and they have to ration it Do managers delegate to machines even when human judgment is better?. Hiring shows this clearly. 70% of hiring managers say AI helps them decide faster, but only 21% of recruiters are very confident their systems don't reject qualified candidates Do hiring managers and job seekers agree on AI fairness?. Meanwhile applicants and filters are escalating against each other, with recruiters spending large parts of their week sorting through spam Are job applicants and employers locked in an escalating AI arms race?. Judgment is the bottleneck in that system, and it is being spread thin instead of paid more.

The most worrying thread is about where future judges come from. Payroll data shows no economy-wide job losses from AI. But young workers in AI-exposed occupations are being hired at 19% lower rates than peers in less-exposed fields, while experienced workers see no comparable gap Is generative AI displacing workers at economy-wide scale?. This fits the warning that unchecked automation erodes the entry-level work where people learn to verify and decide What becomes scarce when AI execution grows cheap?. Firms that are more exposed to AI are also replacing contract workers faster and more cheaply Do firms substitute labor for AI at different rates?. Seniors who already have judgment become more valuable, while the path that produces new ones narrows.

So the answer is yes in principle, but the outcome isn't settled. Brynjolfsson argues that AI built to automate rather than augment shifts bargaining power to whoever owns the technology, and this can lock in an equilibrium where workers lack the leverage to change it Does automating labor concentrate power more than augmenting it?. Accountable judgment only becomes a well-paid resource if firms and institutions choose to pay for it, protect it, and train the next generation of people who have it. The corpus is stronger on the mechanism than on wage data, and it doesn't yet show judgment-heavy roles actually commanding higher pay.