Nobody rewrote your job description. There was no memo, no all-hands announcement, no renegotiated contract. And yet the role you're in — the one with the unchanged title and the unchanged spec — now quietly expects a skill that wasn't on it when you were hired. AI requirements in job postings have roughly tripled since 2022, climbing from about one posting in twenty to nearly one in seven. But the headline isn't the growth. It's where the growth is landing.

The fastest expansion isn't in engineering. It's in marketing, communications, customer service, operations, HR — the functions that spent three years assuming AI was the tech team's problem. In customer service, the share of postings asking for AI skills went from roughly one in three hundred to one in sixteen. Marketing demand tripled. Overall, AI-skill job postings more than doubled in a single year and now account for close to four percent of everything advertised in the US — and the curve is still steepening month by month.

The requirement didn't arrive in your job spec. It arrived in everyone else's.

The Requirement That Sneaks In Sideways

Here's how an expectation gets added to your job without anyone updating your job. Employers aren't rewriting role descriptions in bulk — that would require admitting the roles changed. Instead, the requirement enters through the hiring side: new postings for the same function simply include "AI fluency" or "experience with AI tools" as a line among the others. Your incumbent spec stays the same. The market's spec for your role moves on.

The result is a peculiar kind of gap. You can be fully meeting the requirements of the job you were hired into while quietly falling short of the requirements for the job your function has become. Nobody tells you, because nobody owns the translation: your employer added the expectation to the postings, but nobody's defined what AI competence means inside your actual role. That translation — what does AI fluency look like for a PR professional, a comms lead, an ops manager — has been left as an exercise for the reader.

And the reader, mostly, guesses wrong.

The timing makes the gap sharper than the raw numbers suggest. Requirements that take years to normalise in one part of the economy are being adopted in months inside non-technical functions — because these professions were late to start and are now catching up at compressed speed. A craft that spent 2023 debating whether AI was relevant to it is now posting AI requirements as standard practice in 2026. That's not a gradual shift; it's a correction, and corrections price in fast.

Where the Demand Is Concentrated

The pattern isn't uniform — and that non-uniformity is the most useful part of the data. The functions absorbing the requirement fastest share a common shape: they're communication-heavy, output-dense, and built on producing variations of the same thing at scale. Public relations, marketing, internal comms, customer service, operations. These are exactly the disciplines where the tool's leverage is most direct — not because the work is simple, but because the work already lives inside language, speed and repetition.

Engineering, meanwhile, grew more slowly in relative terms — precisely because it started earlier. The surge you're seeing in non-tech postings isn't tech demand migrating. It's the rest of the economy discovering what engineering discovered three years ago and pricing it into hiring at speed. The lag is the story: non-technical functions are where the repricing is happening right now.

For anyone in one of these functions, the implication is uncomfortable and specific: the market has already decided your craft now includes a tool you may not have learned yet. The requirement isn't ahead of you; it's already posted.

What "AI Skills" Actually Means in a Non-Tech Job

The most common misreading is the assumption that "AI skills" means technical skills — that the requirement is about code, models, prompt syntax, something for engineers to worry about. That's the outdated reading, and it's the one that makes non-technical professionals scroll past the requirement as though it weren't addressed to them.

The labour-market data tells a different story. The demand spreading into non-technical functions isn't demand for people who build AI. It's demand for people who apply it inside a craft that isn't about AI at all. A PR professional who uses AI for media monitoring, sentiment analysis, draft variation and rapid response isn't doing a technical job — they're doing PR with a new toolkit. The skill being priced is applied fluency inside the discipline, not technical competence in the abstract.

Which is why generic familiarity doesn't satisfy the requirement. Employers posting "AI skills" for a communications role aren't asking whether you've opened a chatbot. They're asking whether AI is wired into how you produce the work — the coverage monitoring, the crisis drafting, the pitch tailoring, the reporting. The gap the market is pricing isn't between technical and non-technical people. It's between people who apply AI inside their craft and people who merely know it exists.

Why the Capable Professionals Are the Ones Falling Behind

There's a painful inversion inside this data. The people most exposed to the new requirement aren't the underperformers — they're experienced professionals whose competence was built before the toolkit changed. Their judgment, relationships and craft knowledge are still the asset. But the market has started pricing the craft as incomplete without the applied layer on top of it.

Three failure patterns repeat. First, dismissal: treating the AI line in postings as boilerplate that doesn't really apply — until the shortlist says otherwise. Second, delegation: waiting for the employer to define what AI means for the role, when the data shows employers are adding expectations precisely because they haven't done that defining themselves. Third, generality: learning "AI" as a detached skill — a tool tour, a prompting primer — that never connects to the actual discipline the requirement lives inside.

All three have the same root cause: reading the requirement as a tech story when it's a craft story. The professionals repricing upward right now aren't the ones who learned the most about AI. They're the ones who learned what AI does inside their specific function.

There's a fourth group worth naming separately: the people who genuinely tried, learned the tools, and then found the learning didn't translate. They watched the demos, did the tutorials, could talk fluently about what AI can do — and still produced Monday's work the slow way. That's not a learning failure. It's what happens when the skill is taught as a subject rather than as a layer inside the work. General knowledge about AI doesn't change output. Applied fluency inside the craft does. The requirement in the postings is for the second, and the market has stopped paying for the first.

The Advantage of Being Inside the Craft

Here's what makes this moment genuinely favourable for non-technical professionals, if they move: domain expertise is the scarce half of the equation. AI fluency can be learned far faster than a decade of PR judgment can be taught. The market isn't looking for technologists to take over communications. It's looking for communicators who've added the layer — and there are still relatively few of them, which is exactly why postings have had to start asking.

For anyone in public relations specifically, this is the sharpest version of the trend: media monitoring, sentiment tracking, crisis response speed, content variation at scale — the discipline is one of the clearest cases where applied AI multiplies the craft rather than replacing it. The professionals who wire it in now aren't just meeting the new requirement. They're the ones defining what it will mean.

And the compounding advantage goes to whoever moves earliest inside their own function. The first PR professional in an agency to wire AI into the monitoring workflow doesn't just keep their role current — they become the person who defines how the team uses it, which is a different job entirely. Late adopters inherit a standard. Early adopters write it. In a repricing market, that's the difference between meeting a requirement and owning the new shape of the role.

Our AI for Public Relations course teaches exactly that application layer — a daily-email course on using AI inside the actual workflows of PR: monitoring, drafting, analysis and response, built around the craft rather than around the tool. If you're in a different function, the same applied logic runs through our AI Automation course and the wider catalogue.

The requirement is already in the postings for your function — tripled since 2022, still climbing. The only open question is whether it's in your skillset when the next version of your job gets advertised. The window where applied AI was a differentiator is closing; it's becoming the baseline. Baselines don't wait for invitations.