Zara Reyes: Megan, hey — I spent twenty minutes this morning just scrolling job postings because I could not stop.
Megan Skiendel: Scrolling job postings. On purpose.
Zara Reyes: On purpose. Because Indeed's Hiring Lab dropped a report July 13th and the number — 63% of U.S. job postings with 'AI' in the title are outside traditional tech companies. Not inside. Outside. And I could not stop clicking.
Megan Skiendel: Wait — outside tech is the majority?
Zara Reyes: Majority. Sales, HR, legal, nursing administration — 822 distinct AI-touched job titles in Q1 2026 alone. That's what we're getting into today: whether any of that is real or whether it's the most expensive LinkedIn performance art we've ever seen.
Megan Skiendel: Honestly, 'AI-touched.' That phrasing is doing a lot of work.
Zara Reyes: That's the whole take. Those titles tripled since 2022, postings jumped 173% year-over-year in Q1 — and meanwhile recruiters are staring at tens of thousands of unfilled nursing and teaching slots. So the question this whole episode is trying to answer: is AI hiring an actual labor market shift, or is it theater?
Megan Skiendel: And I'm guessing your answer is theater. With very good lighting.
Zara Reyes: Okay, theater with good lighting — I'll stand there. But Pawel Adrjan at Indeed's Hiring Lab is pushing back on exactly that read.
Megan Skiendel: Adrjan's point being — what, the spread is real?
Zara Reyes: That employers are integrating AI skills into roles that had nothing to do with software or data science before. Not just slapping a word on a posting. In five of the six countries Indeed's Hiring Lab tracked — five — more than half the AI-touched titles were outside tech. That's not a branding blip.
Megan Skiendel: No, here's where I pump the brakes. Think about what happened when factories got electric motors in the 1910s. They didn't just rename 'steam operator' to 'electric operator' and go home. The floor plan changed. The shift schedule changed. Entirely new coordination roles appeared. The work reorganized around the new thing. That's actually what the Ramp data is pointing at — companies that adopted AI grew headcount ten percent over two years, entry-level hiring up twelve percent. That is a Jevons Paradox situation. William Stanley Jevons figured this out in the eighteen-hundreds watching coal: make a resource more efficient, and consumption expands, it doesn't shrink. So the question isn't just 'is the job title real.' It's whether the underlying work genuinely reorganized.
Zara Reyes: Entry-level up twelve percent — honestly, that number surprised me. That runs against basically everything the layoff narrative said was happening.
Megan Skiendel: It does. But — and this is the honest answer — we don't know the split. Of those 822 AI-touched titles, how many require you to actually do something new versus just understand that ChatGPT is in your workflow? The sourcing is thin. I'd want to see what Kyndryl and IBM are flagging about workforce readiness before I call it a full reorganization.
Zara Reyes: Kyndryl's whole finding is that adoption is outpacing readiness — so the roles are posting faster than companies can actually staff them with people who know what they're doing.
Megan Skiendel: Which is either proof the work is real and genuinely new, or proof that employers are posting aspirationally. Both things produce the same job posting.
Zara Reyes: No but — that's the cleanest version of the uncertainty. It's probably both. And we don't know the ratio.
Megan Skiendel: And that ratio problem is exactly where your hot take earns its money. Because the nursing shortage — honestly, that's not a ratio question. That's a choice.
Zara Reyes: It's a choice. Full stop.
Megan Skiendel: Think about what a hospital recruiter is actually doing right now. Forty-seven open RN slots — real vacancies, patients waiting. She's got a posting budget. LinkedIn's algorithm tells her AI engineer is the number one fastest-growing U.S. job title in 2026. The cultural gravity in that interface pulls her one direction. The care crisis pulls the other. She's not confused. The tool is just — it's pointed somewhere else.
Zara Reyes: And the nursing shortage predates AI by decades. We didn't deprioritize it because suddenly there were shinier jobs. We deprioritized it because it pays badly and the relational labor it requires — AI is not the reason it's unglamorous.
Megan Skiendel: No, that's right. And the IMF is actually saying this out loud. Florence Jaumotte and her colleagues — SDN/2026/001, January 2026 — they called the mismatch a structural policy problem. Not a self-correcting market outcome. A choice that requires intervention.
Zara Reyes: Wait, the IMF is saying the market won't fix this on its own?
Megan Skiendel: Explicitly. And that matters because — look, Sam Altman in Sydney saying he's 'delighted to be wrong' about categories of work being totally gone? Fine. His reversal doesn't close a single one of those forty-seven RN slots. The displacement didn't happen the way he predicted, but the shortage is still sitting there, fully intact.
Zara Reyes: And the 9 in 10 HR leaders who regret the AI-driven layoffs — that's the same instinct run backwards. Panic in, panic out. They cut, now they can't rehire fast enough, and the people they need most are in fields where the pipeline was already broken before any of this started.
Megan Skiendel: Nine in ten. That is — frankly, that number should be the headline and it's not.
Zara Reyes: It's not, because the prestige story is more legible. And — actually, the part that makes all of this worse? The Kyndryl data on workforce readiness, and what IBM is flagging about the skills gap — that's where we find out this isn't a paradox and it isn't pure theater. It's a bifurcation, and organizations are failing on both sides simultaneously. We'll get there.
Megan Skiendel: Both sides simultaneously — and that's the part that doesn't resolve cleanly. Because Kyndryl isn't saying adoption is slow. They're saying it's outrunning readiness. Which means companies are mid-deployment with no one who actually knows how to run the thing.
Zara Reyes: And IBM confirms the same direction — demand for AI-skilled workers outpacing supply, even with AI engineer postings up 143% year-over-year in 2025. You cannot hire your way out of a gap that grows faster than the hiring.
Megan Skiendel: Right — but the part that doesn't fit the clean story is the layoff regret data sitting right next to it. Nine in ten HR leaders regretting those cuts. That's not — I mean, those are two failures happening at the exact same time. You cut the people you had, you can't staff the AI roles you need, and the workers you still have aren't trained for what you deployed.
Zara Reyes: That's not a paradox.
Megan Skiendel: No. That's a choice architecture. Organizations decided to move fast on the tech and slow on the people. Kyndryl's making that concrete — this is an enterprise-wide readiness failure, not a few laggard companies.
Zara Reyes: So what's the defensible claim once you strip the hype? Because I want to be precise about this — it's not 'AI is eating jobs,' Altman already walked that back in Sydney. But it's also not 'AI lifts all boats.'
Megan Skiendel: The defensible claim is: we are running two crises in parallel and treating neither with urgency. An AI skills gap at the high end — IBM, Kyndryl, the 143% posting surge that still can't fill demand. And a structural staffing collapse in nursing, teaching, trades — which predates all of this. And the HR regret data is the hinge. It proves organizations already paid for prioritizing the signal over the substance.
Zara Reyes: Lowkey that's the whole episode. The theater framing gets you halfway. But the Kyndryl and IBM data make it concrete — this isn't theater, it's a simultaneous misallocation on both ends.
Megan Skiendel: And nobody is running late on that. They're already behind.
Zara Reyes: I'll half-concede here — theater was maybe too clean. Some of those 822 AI-touched titles are doing genuinely new work. I'll give that. But if eighteen months from now nursing vacancies are still in the tens of thousands and AI-titled postings hit a thousand distinct titles, we can't call that a market failure anymore. That's a preference. That's what the silence in the Indeed's Hiring Lab data actually says.
Megan Skiendel: The labor market isn't broken. That's — honestly, that's what I'd want someone to take away. It's working exactly as designed. We just haven't been willing to look at who designed it, and for whom.
Zara Reyes: That's a hard line to sit with. Good one to end on, though.
Megan Skiendel: Frankly, the best ones usually are.
