Cleo Rios: Mia, hey — quick question before we get into it: if someone handed you a list of fifteen jobs and said 'AI will never touch these,' what's your first instinct?
Mia Drake: Honestly? Immediate suspicion. Who made the list and what are they selling?
Cleo Rios: Correct instinct, wrong conclusion — because this one is actually built on something. Bryan Robinson, Ph.D., publishes it in Forbes on August 16, 2026, and the data backbone is a Resume Genius report that mapped occupations by automation exposure, salary, and growth rate. It's not vibes. It's methodology. And some of the findings are genuinely — I mean, Wind Turbine Service Technicians. Six figures. AI-resistant because you are physically three hundred feet up in an unstructured environment doing manual diagnostics. Thirteen percent projected growth.
Mia Drake: Okay the six figures part — that's not what I expected from wind turbines.
Cleo Rios: Nobody expects it! That's the whole thing. And then you look at Nurse Practitioners — hundred and thirty-five thousand, eight hundred and eighty dollars median, twenty percent projected growth — and the resistance argument there is totally different. It's not physical danger, it's relational judgment. Empathy as a diagnostic tool. Resume Genius actually puts those in separate categories within their taxonomy, which — I find that distinction genuinely useful, and also genuinely fragile.
Mia Drake: Fragile how? Like the categories don't hold under pressure?
Cleo Rios: Like — the list is a snapshot. And a snapshot published in 2026 assumes the AI landscape stays roughly where it is while someone spends two years getting a Nurse Practitioner credential. That assumption is, I think, doing a lot of heavy lifting that nobody's talking about.
Mia Drake: So the real question is whether 'AI-resistant' is a fixed property of a job or something that expires. That's what we're actually asking.
Cleo Rios: Right — and that's exactly where the Resume Genius taxonomy gets interesting, because they're not treating 'AI-resistant' like it's one thing. They actually split it into three buckets. And the logic inside each bucket is different.
Mia Drake: Okay, walk me through the buckets.
Cleo Rios: Think of it like this — bucket one is people who work WITH AI. Software developers, data scientists. They're not hiding from it, they're the ones holding the wheel. Bucket two is people whose mistakes are too expensive for a machine to own. Financial managers, actuaries — like, if an actuary miscalculates mortality risk, someone's retirement evaporates. You need a human signature on that. And bucket three is people whose job literally requires a body in a room. Healthcare workers, skilled tradespeople. That's it. That's the whole frame.
Mia Drake: So where does the wind turbine technician land? I'm guessing bucket three?
Cleo Rios: Hard bucket three. And here's the actual mechanism — it's not just 'physical job, therefore safe.' It's that the environment is unstructured. A robot can be trained on a fixed factory floor. But diagnosing a gearbox failure at two hundred and eighty feet in crosswind, with ice on the rungs, different turbine model than last week? That variability is the barrier. Resume Genius flags this specifically for Wind Turbine Service Technicians AND Industrial Machinery Mechanics — the environment never repeats itself exactly.
Mia Drake: Oh — so it's not the height, it's the chaos.
Cleo Rios: The chaos, yes! Which is also, I mean — that's why Physical Therapists at a hundred and two thousand dollars with eleven percent growth and Occupational Therapists at a hundred thousand, fourteen percent growth, both land in that bucket. Every patient's body is a different unstructured environment. A shoulder after a car accident isn't the same problem twice.
Mia Drake: And McKinsey's own 2025 survey puts automation at nine percent of global work hours by 2030 — so even their projection assumes most of this stays human.
Cleo Rios: Nine percent. Which sounds almost reassuring until you remember that nine percent hits some workers at a hundred percent. But the taxonomy — okay, the taxonomy at least tells you which direction to move.
Mia Drake: But that taxonomy — it's built for a world where AI takes a task, not a world where AI takes the whole Tuesday.
Cleo Rios: Okay yes — that is exactly the thing. Picture a data scientist, nine a.m., coffee, reviewing model outputs. Every individual task — she's evaluating outputs, she's deciding what to flag, she's writing the stakeholder summary — each one reads as human-critical. But Ravish Gupta and Saket Kumar published a paper in 2026 that asks a completely different question: can an agentic AI system run the entire sequence? Data ingestion, anomaly detection, interpretation, report delivered to the executive deck — no her. That's the workflow-level question. And the Acemoglu-Restrepo framework, which everyone was using for task-level risk, had to be extended specifically to account for this.
Mia Drake: Wait — Forbes literally lists software developers and data scientists as safe. They're in the AI Collaborators bucket.
Cleo Rios: Yes. And that is the problem. Because 'AI Collaborator' assumes she's always in the loop — that the workflow needs her to proceed. Gupta and Kumar's whole argument is that agentic AI, and I mean autonomous systems doing multi-step reasoning, invoking tools, making decisions without prompting — that class of system doesn't substitute for her task, it makes her workflow optional.
Mia Drake: So the category looks safe but the job is being hollowed from underneath.
Cleo Rios: Hollowed is exactly the word. And what's wild is — the prior task-level analyses, the ones that flagged office and admin roles at up to ninety-six percent automation exposure, those were already alarming. But they were still playing by task-substitution rules. Gupta and Kumar say: no, agentic systems don't play by those rules. They execute end-to-end. Which means a job category can score relatively safe at the task level and still get wiped at the workflow level. The Forbes framework doesn't account for that distinction at all.
Mia Drake: No way. So Resume Genius is measuring the wrong unit.
Cleo Rios: It measured the right unit for 2023 AI. It's measuring the wrong unit for 2026 AI. And — wait, actually this is the part that makes it worse — even if the map is accurate today, it only reaches the workers who can afford the credential ladders. And the highest-paying safe roles cluster around exactly the degrees that displaced workers can't easily access. We'll get there.
Mia Drake: I don't buy that the AI Collaborator category survives that at all then.
Cleo Rios: And that's the map problem — even if every salary figure on the Forbes list is BLS-accurate, the map only works if you can reach the territory. Nurse Practitioners at a hundred thirty-five thousand, eight hundred and eighty dollars, twenty percent growth — that's real. But that credential is two-plus years of graduate school with prerequisites that a fifty-two-year-old data entry supervisor in a rural county probably hasn't touched since, I don't know, 2001.
Mia Drake: And no nursing school within commuting distance.
Cleo Rios: Exactly. The list exists for her. It just — it doesn't reach her. And what's maddening is Resume Genius puts Wind Turbine Service Technicians on the same list, same document, no advanced degree required, real six-figure growth trajectory — and that actually IS accessible. That's the crack in the credential wall.
Mia Drake: So wind turbines are the outlier that breaks the whole framing open — but how narrow is that crack, actually?
Cleo Rios: Narrow. I mean — Wind Turbine Technicians, Industrial Machinery Mechanics, that tier exists. But it's a thin band on the list. The credential asymmetry is that those accessible roles and the hundred-and-thirty-five-thousand-dollar roles get presented as one coherent menu, which — no. Absolutely not. That's not one menu.
Mia Drake: It's like putting a studio apartment and a brownstone on the same real estate flyer and calling it 'options.'
Cleo Rios: Yes! And the sources — Bryan Robinson's Forbes piece, the Resume Genius report underneath it — they don't address retraining pathways at all. There's no infrastructure question. No 'how does the displaced administrative worker actually get here.' The BLS growth projections are real, the salary floors are real, and the transition mechanism is just... absent.
Mia Drake: So the framework glosses over the whole transition question — which is the only question that matters if you actually need to move.
Cleo Rios: The whole question. And that's not a footnote, that IS the equity problem baked into the list.
Mia Drake: The thing that actually settled for me — Resume Genius's methodology for why a job stays resistant? It's not fully documented anywhere. Like, the durability of those designations is just... unverified. We're making multi-year decisions on a snapshot that doesn't even show its own expiration date.
Cleo Rios: And the Acemoglu-Restrepo framework — the thing everyone was using — had to be extended in 2026 specifically because it underestimated agentic AI. The model was wrong and needed fixing mid-game. That's not a footnote. That's the whole situation.
Mia Drake: McKinsey says nine percent. Someone else says ninety-six percent exposure for specific roles. Both are technically describing real things and somehow they're supposed to reach the same worker making the same decision.
Cleo Rios: I mean — we're not saying don't use the map. The map is real. It's just got no printing date on it.
Mia Drake: And the wind turbine tech climbing the tower today is probably safer than the data scientist who assumed complexity was protection. That's the honest place this lands.
