AI made hard skills cheap. The expensive skills now are human.

Run back through this series and one pattern repeats. Engineers watched AI absorb code production and got repriced on architecture. Sellers watched it absorb outreach volume and got repriced on discovery and deal strategy. Designers watched it absorb pixels and got repriced on taste. Different jobs, identical trade: the cost of producing a competent technical draft fell toward zero, everywhere, at once.

Markets reprice both sides of a trade like that. When production gets cheap, the inputs production can't supply get expensive: knowing which problem is worth solving, judging whether the output is actually good, earning the trust that makes someone act on it, explaining it to a person who can say yes, and being willing to look at a confident draft and call it wrong. None of that comes out of a model. All of it now decides who gets paid.

The hiring data agrees. The World Economic Forum's Future of Jobs Report 2025 expects 39% of workers' core skills to change by 2030, and the skills employers rank as rising fastest read remarkably human: analytical thinking, resilience and flexibility, leadership and social influence, and creative thinking, sitting right alongside AI and big data. Read that list again. The human skills and the machine skills rise together, on the same page, and that pairing is the real headline.

We called technical skills 'hard' because a test could certify them and a resume could list them. Judgment, communication, and trust could only be observed over time, so the market priced what it could measure and waved at the rest. 'Soft' was an accounting limitation, not a difficulty rating. Anyone who has tried to run a hard conversation, win over a skeptical committee, or tell a senior stakeholder their favorite idea fails knows which skills were actually hard.

Now the measurable half is cheap and the unmeasurable half is the product. LinkedIn found that communication topped its list of the most in-demand skills, even amid the AI hiring boom. That ranking makes sense once you see what changed: when machines write the first draft of everything, the scarce person is the one who can decide what the draft should argue, fix where it's wrong, and deliver it so a human being acts on it.

Here's the nuance that keeps this from becoming a comfortable excuse. Human skills only command the premium when they're paired with AI fluency. Judgment with no production speed loses, consistently, to judgment with it, because AI amplifies whoever wields it and an empty production tier amplifies nothing. The market isn't paying for human skills instead of AI skills. It's paying for the pairing, and the pairing is the position. The skills worth building are the ones that close that loop: brief the machine, judge the output, ship with your name on it.