Silicon Valley Has the Highest AI Job Risk in America | Tuft's New Data Is Wild

Tufts University just released the first-ever American AI Jobs Risk Index (March 24, 2026) — and the results are not what most people expected. Silicon Valley has the highest AI job risk of any U.S. city at 9.9%. Boston, D.C., Seattle, New York — the Wired Belts — are the most exposed. Writers face 57% displacement risk. Computer programmers 55%. And the day after this report dropped, ADP published a survey of 39,000 workers: only 22% feel their job is safe. On this week's bonus episode, Carlo Thompson breaks down every number, explains what the geographic story means for your career, and gives you four specific actions to take this week.

📊 Sources: Tufts Digital Planet (March 2026), ADP Research Today at Work 2026, Fortune/Duke CFO Survey

👂 Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/surviving-ai-job-automation-workforce-future-insights/id1864360631
▶️ Subscribe on YouTube: https://www.youtube.com/@SurvivingAIRisk/video

Episode resources

https://docs.google.com/document/d/12w6mzGndqtqx4tlHFxaQ2lu80bHIcnxMgIIWo5ltZXs/edit?usp=sharing

American AI jobs risk index 2026, Tufts Digital Planet AI jobs, Wired Belts Rust Belts AI, AI job displacement by city, Silicon Valley AI risk, ADP job security survey 2026, CFO AI layoffs confession, writer programmer AI risk, AI automation geography, 9.3 million jobs AI risk

Transcript

SPEAKER_01 0:00

Welcome back to Surviving AI with Carlo Thompson. This is your bonus Wednesday deep dive where we take the most important, most impactful news of the week and really break it down for your career.

SPEAKER_00 0:12

And I mean, this week is a huge one. We have two reports that drop back to back that every single person in this audience needs to hear about.

SPEAKER_01 0:19

Yeah, absolutely. And we're not talking about preparing for it a year from now or you know waiting until this hits mainstream cable news. We're talking about right now.

SPEAKER_00 0:26

Aaron Powell Right. The data is already here.

SPEAKER_01 0:28

Exactly.

SPEAKER_00 0:28

Artificial system online.

SPEAKER_01 0:41

So on March 24th, Tufts University released what they're calling the first ever American AI Jobs Risk Index. It's comprehensive, it's entirely data-driven, and it names like specifically names which occupations, industries, and cities are most at risk from AI displacement.

SPEAKER_00 1:00

And then literally the very next day, March 25th, ADP published its 2026 Global Workforce Survey. They went out and asked 39,000 workers across 36 countries, just one simple question.

SPEAKER_01 1:10

Aaron Powell Right. They asked, do you feel your job is safe?

SPEAKER_00 1:13

Yeah. And that answer it might honestly be the most important number in workforce research right now.

SPEAKER_01 1:18

Aaron Powell We're going to spend some time on both of these today because look, this is actionable intelligence. Previously, AI job predictions were like these vague national weather forecasts, right?

SPEAKER_00 1:29

Trevor Burrus So totally. Like someone telling you, you know, it's going to rain globally over the next decade. Trevor Burrus, Jr.

SPEAKER_01 1:34

Right. And you're just looking out your window wondering, well, do I need an umbrella today or not? But this Tufts index, it's like a hyper-local Doppler radar. Trevor Burrus, Jr.

SPEAKER_00 1:42

That's a great way to put it. It shows exactly whose house is about to get struck by lightning.

SPEAKER_01 1:47

Aaron Powell, so let's get into it. We really need to move away from these future sci-fi projections and look at the reality of right now. Before we even talk about who is losing their jobs, we have to establish the sheer scale of what AI can actually do today.

SPEAKER_00 2:02

Aaron Powell Yeah, let's look at the headline numbers. Tufts estimates that 9.3 million US jobs are at risk of displacement in the next two to five years.

SPEAKER_01 2:11

Wow. 9.3 million.

SPEAKER_00 2:14

And that's just the midpoint, right? The range actually spans from 2.7 million on the low end all the way up to 19.5 million on the high end.

SPEAKER_01 2:21

That is a massive range. What does that look like financially?

SPEAKER_00 2:25

Financially, we are looking at$757 billion in annual wages exposed at that midpoint. And if we hit the high end of their projection, it's$1.5 trillion in annual income at risk.

SPEAKER_01 2:38

Aaron Powell Okay, wait, hold on. I have to push back here a little bit. Because hasn't every major technological revolution I mean, from the steam engine to the spreadsheet, hasn't it always sparked this exact panic that millions of jobs would disappear?

SPEAKER_00 2:52

Sure, yeah. The historical argument.

SPEAKER_01 2:53

Right. Like historically, we've always just created new, better jobs. When the ATM was invented in the 80s, everyone thought bank tellers were going completely extinct. But banks just opened more branches and hired more tellers. So why is this 9.3 million number any different than the ATM panic?

SPEAKER_00 3:08

Aaron Powell Well, because the ATM automated a single mechanical task, just handing out cash.

SPEAKER_01 3:14

Okay.

SPEAKER_00 3:14

It didn't automate the cognitive reasoning required to, you know, cross-sell a mortgage or resolve a really complex account dispute. What we are facing right now is cognitive automation. Ah and the crucial definition we need to establish from this Tufts data is what at risk actually means in this context.

SPEAKER_01 3:31

Which is what?

SPEAKER_00 3:32

Cuffs does not mean AI is going to learn how to do these jobs in five years. They mean AI can perform the primary functions of these jobs right now. Today, the technology already exists on the server.

SPEAKER_01 3:45

Wait, really? So if the technology is already capable, what accounts for that massive variance? I mean, the gap between 2.7 million and 19.5 million jobs is huge. What's holding it back?

SPEAKER_00 3:57

Honestly, pure corporate bureaucracy.

SPEAKER_01 3:59

You're kidding.

SPEAKER_00 4:00

No, seriously. The only thing saving those millions of jobs right now is just the sheer friction of human organizations. I mean, think about it. It takes time for procurement departments to approve new software vendors. Right. It takes time for HR to reorganize reporting structures or to rewrite company policies around data security. The bottleneck is not the AI getting smarter, the bottleneck is how fast corporate leaders can actually deploy the tools they already have.

SPEAKER_01 4:25

So you're saying the technology isn't the barrier anymore? The only shield protecting a worker's income right now is basically how slowly their HR department moves.

SPEAKER_00 4:33

Precisely. And look, that friction isn't going to last. Once a competitor figures out how to streamline those processes and deploy the AI, the rest of the market will be forced to accelerate just to survive.

SPEAKER_01 4:46

But okay, a$1.5 trillion wage exposure isn't just going to be spread evenly across the country like peanut butter, right? It's going to pool somewhere.

SPEAKER_00 4:55

Oh, absolutely.

SPEAKER_01 4:56

Because looking at the geographic data from the report, there is this massive, highly ironic twist about where ground zero actually is.

SPEAKER_00 5:06

Yeah. Tufts named this section of the report the wired belts are the new rust belts. Trevor Burrus, Jr.

SPEAKER_01 5:11

The wired belts.

SPEAKER_00 5:12

Right. Think about the old rust belt. Cities like Detroit, Cleveland, Pittsburgh. Those are the geographic centers of American economic power in the mid-20th century. Trevor Burrus, Jr.

SPEAKER_01 5:21

And they were totally decimated when physical manufacturing got automated or pushed offshore. Trevor Burrus, Jr.

SPEAKER_00 5:26

Exactly. So the new rust belt is the wired belt. We are talking about the knowledge capitals. The highest risk is concentrated in the cities that drive the digital economy.

SPEAKER_01 5:34

So Siliton Valley.

SPEAKER_00 5:35

Number one in the nation. Specifically the San Jose metro area. 9.9% of all jobs in that region are at risk right now.

SPEAKER_01 5:42

Wait, let me get this straight. The people who spent the last decade building the AI in San Jose are the most at risk from the AI.

SPEAKER_00 5:49

It's the defining paradox of this whole economic shift.

SPEAKER_01 5:52

The creators built the exact tools to automate themselves.

SPEAKER_00 5:56

Yeah. Because look, these cities have the most educated, highest income populations in America, but their entire localized economies are built on digital processing, text generation, coding, and data analysis. Trevor Burrus, Jr.

SPEAKER_01 6:09

Right. It's all screenwork.

SPEAKER_00 6:10

Exactly.

SPEAKER_01 6:11

Yeah.

SPEAKER_00 6:11

Because their output is entirely digital, it maps perfectly onto the core capabilities of current AI. A factory worker's output is physical, right? A Silicon Valley worker's output is code and text, and AI is exceptionally good at code and text.

SPEAKER_01 6:26

It's a dark kind of poetry, honestly. But the tech hubs are only part of the story. The data on university towns completely caught me off guard.

SPEAKER_00 6:32

Oh, the vulnerability of college towns is one of the most surprising findings in the whole index. Metros like Durham Chapel Hill, Boulder, Ann Arbor, Ithaca, Madison. They're all sitting in the top 25 most vulnerable regions in the country.

SPEAKER_01 6:50

But why would academia be on the chopping block? I mean, you can't replace a tenured professor running a physical chemistry lab with a chatbot.

SPEAKER_00 6:57

You can't, but you're only looking at the tip of the iceberg there. The vulnerability isn't the professor giving the lecture. It's the massive unseen infrastructure of academic research support beneath them.

SPEAKER_01 7:08

Oh, the administration.

SPEAKER_00 7:09

Yes. Modern universities employ absolute armies of knowledge workers whose primary job is administrative bloat and grant writing.

SPEAKER_01 7:16

Right.

SPEAKER_00 7:17

Think about how a university gets funding. It requires assembling these 50-page grant proposals, synthesizing massive literature reviews, formatting documents to incredibly strict government standards.

SPEAKER_01 7:27

Which is exactly what a large language model is designed to do.

SPEAKER_00 7:30

It is an LLM's dream task, taking vast amounts of unstructured text, identifying the key themes, and outputting a highly structured bureaucratic document.

SPEAKER_01 7:39

Aaron Powell So the people doing that work in Boulder and Ann Arbor are facing immediate displacement risk just because their daily output is essentially what sophisticated data formatting.

SPEAKER_00 7:48

Aaron Powell Exactly. That's all it is.

SPEAKER_01 7:50

Wow. Okay, so if the tech hubs and the university towns are the new rust belt, what geography is actually safe? Where are the havens?

SPEAKER_00 7:58

Aaron Powell The safest havens are the trades heavy and services heavy metros. Cities with high concentrations of physical, unpredictable work.

SPEAKER_01 8:06

Aaron Powell Like plumbing and electrical.

SPEAKER_00 8:08

Yeah. Plumbing, direct healthcare delivery, electricians, HVAC repair. AI cannot navigate the chaotic physical environment of a flooded basement to solder a pipe.

SPEAKER_01 8:18

Right, it doesn't have hands.

SPEAKER_00 8:20

Exactly. And it can't physically lift and turn a patient in a hospital bed. Automation requires predictability. Physical trades operate in environments of constant, unpredictable variables.

SPEAKER_01 8:30

So zooming in from the city level, we really need to look at the specific desks inside those wired-built office buildings. Who exactly is doing the vulnerable work? What's the occupation hit list?

SPEAKER_00 8:40

The occupation level percentages are genuinely jarring. Writers and authors face a 57% displacement risk. It means nearly six in ten of their daily primary functions can be handed off to AI today. Computer programmers are at 55%. Web and digital interface designers are also at 55%.

SPEAKER_01 9:00

What about broader industries?

SPEAKER_00 9:01

Aaron Powell Looking at entire industries, the information sector sits at the very top with an 18% risk level. That's followed by finance and insurance at 16%, and professional, scientific, and technical services also at 16%.

SPEAKER_01 9:14

I really want to dig into that 55% risk for programmers because it feels like entry-level coders spent years building the exact shovel that's digging their own career grave. If an AI can write boilerplate code, what is actually left for the junior developer?

SPEAKER_00 9:30

Aaron Powell Well, think about what a junior programmer actually does on like a random Tuesday morning. A massive chunk of their day is not designing complex innovative systems.

SPEAKER_01 9:39

Right. They're not architects yet.

SPEAKER_00 9:41

No, it's routine digital bricklaying. It's writing boilerplate code to connect one API to another. It's spending three hours hunting down a missing semicolon or debugging some minor feature. Trevor Burrus, Jr.

SPEAKER_01 9:51

Which an AI can do instantly.

SPEAKER_00 9:53

Exactly. An AI can read an entire company's code base and write that API integration in four seconds. That routine entry-level work is the 55% that is vanishing.

SPEAKER_01 10:04

Which means the traditional on-ramp to becoming a senior developer is effectively being destroyed. You can't become a master architect if there are no apprenticeships left for laying bricks.

SPEAKER_00 10:13

Yes. That is the long-term structural threat to the tech industry. The programming jobs that survive will require architectural judgment, complex system design, deep client communication. Yeah. But that entry-level ramp is critically exposed.

SPEAKER_01 10:27

Meanwhile, as white-collar knowledge workers face this massive exposure, there is a literal blue-collar boom happening in the background to support all of this.

SPEAKER_00 10:35

Oh, the irony deepens even further here. To run the AI models that are replacing the data analysts of the junior coders, you need massive hyperscale data centers.

SPEAKER_01 10:44

Right. The cloud isn't actually in the sky, it's a building.

SPEAKER_00 10:47

Exactly. And do you know who builds and maintains those buildings? Pipe fitters to install the cooling systems, HEAC technicians to manage the thermal load, ironworkers to frame the buildings, electricians to wire the server racks.

SPEAKER_01 11:00

So AI is actively creating unprecedented blue-collar demand.

SPEAKER_00 11:05

Yes, because the digital cloud requires an enormous footprint in the physical world.

SPEAKER_01 11:10

Okay, so we have the hard data on what jobs are exposed, we know the what and the where. Now we shift to how the human beings sitting in those jobs are reacting, and more importantly, what the executives above them are confidentially planning.

SPEAKER_00 11:24

This is where that ADP 2026 Global Workforce Survey really provides the human context. So they surveyed 39,000 workers across 36 countries.

SPEAKER_01 11:34

That's a huge sample size.

SPEAKER_00 11:36

It's massive. And they asked a very direct question. Do you feel your job is safe from elimination? And only 22% strongly agreed that their job is safe.

SPEAKER_01 11:46

Only 22%. That means 78% of the global workforce is living with active daily job anxiety.

SPEAKER_00 11:52

Yes. Nearly eight in ten workers are basically waiting for the other shoe to drop. But what's truly revealing is how that data fractures when you break it down by management level. It creates a phenomenon we can call the ladder of ignorance.

SPEAKER_01 12:04

The ladder of ignorance. Explain the mechanics of that. How does that work?

SPEAKER_00 12:07

So confidence in job safety actually increases the further up the corporate hierarchy you travel. Because climbing the ladder distances you from the actual work being done.

SPEAKER_01 12:17

Oh wow.

SPEAKER_00 12:18

The individual contributors, the 24-year-old financial analysts building the spreadsheets, the junior developers writing the Python scripts, the marketing associates drafting copying, they feel the least safe. Only 18% of them feel secure.

SPEAKER_01 12:32

Because they're the ones sitting at their desks watching Chat GPT format an entire quarterly earnings report in six seconds. They see the automation happening in real time.

SPEAKER_00 12:41

Exactly. They are staring the replacement technology right in the face. But as you climb the ladder, the perception shifts. Frontline managers are slightly more confident. Middle managers are even more secure.

SPEAKER_01 12:53

And the executives.

SPEAKER_00 12:54

By the time you reach the C-suite, 35% of them feel safe. The CEO sitting five levels up just receives a clean, polished PowerPoint presentation. They don't see the AI doing the heavy lifting. They just assume their team is working really efficiently.

SPEAKER_01 13:08

So the people sitting in the strategy meetings are the most confident because they are entirely detached from the tectonic shift happening on the ground floor.

SPEAKER_00 13:15

Exactly.

SPEAKER_01 13:16

But wait, those executives aren't just sitting there feeling blindly confident. They're actively making moves behind the scenes. And this brings us to the Duke University CFO survey. I mean, this piece of data is the absolute smoking gun of this entire deep dive.

SPEAKER_00 13:32

It absolutely is. When Duke surveyed these chief financial officers privately, like in a confidential academic setting where they didn't have to worry about panicking Wall Street or protecting their stock prices or facing internal PR backlashes, the CFOs admitted something huge.

SPEAKER_01 13:47

What did they say?

SPEAKER_00 13:48

They admitted that their planned AI-related job cuts for 2026 are nine times higher than what they have publicly disclosed to the market.

SPEAKER_01 13:56

Nine times higher. Let that land for a second. Nine times higher.

SPEAKER_00 13:59

It's staggering. Furthermore, 44% of CFOs plan AI-related headcount reductions this year alone. Nearly half of all major U.S. companies are preparing to cut jobs within the next few months.

SPEAKER_01 14:11

So basically, the C-suite is publicly downplaying layoffs to protect their stock prices, but confidentially admitting to a bloodbath. They're leaving the individual contributors who already sense the danger to shoulder all the anxiety in the dark. Yes. You cannot operate your career on vibes when the executives above you are operating on hidden spreadsheets.

SPEAKER_00 14:30

And we really have to look at the underlying economic theory driving those hidden spreadsheets too. It's a concept known as Solo's Paradox.

SPEAKER_01 14:38

I'm not familiar with Solos Paradox. What is the mechanism behind that?

SPEAKER_00 14:41

So the paradox was first identified in the late 1980s by an economist named Robert Solo. He famously said, you can see the computer age everywhere except in the productivity statistics. During the dawn of the personal computer, companies bought machines for everyone, expecting massive immediate efficiency gains. But productivity barely moved.

SPEAKER_01 15:02

Why?

SPEAKER_00 15:03

The mechanism behind the paradox is friction. It takes years to train workers on new tools like Lotus 123 back then. It takes years to restructure legacy workflows. Moreover, new technology often creates new shadow work.

SPEAKER_01 15:17

How does that apply to AI today then?

SPEAKER_00 15:19

Executives are cutting jobs right now based on assumed massive productivity gains. They believe giving their team an enterprise AI tool will make the remaining workforce infinitely more efficient.

SPEAKER_01 15:32

But it isn't.

SPEAKER_00 15:33

Well, the macroeconomic data currently shows a very modest 1.8% productivity gain from AI so far.

SPEAKER_01 15:39

Wait, if the productivity gain is only 1.8%, why are they cutting jobs nine times faster?

SPEAKER_00 15:45

Because the AI is generating new shadow work. Let's say a worker uses an LLM to generate a massive detailed email in two seconds. The recipient now has to spend five minutes reading an AI-generated email to ensure there are no hallucinations or errors.

SPEAKER_01 16:00

Ah, so the time saved on generation is lost on verification.

SPEAKER_00 16:04

Exactly. The workflows have not been optimized yet. But here is the really malicious part for the worker. The paradox does not matter to your employment status.

SPEAKER_01 16:12

What do you mean?

SPEAKER_00 16:12

Whether the AI actually delivers a 20% gain or 1.8% gain, the CFO has already made the decision to reduce headcount. The cuts are happening based on the executive's belief in future productivity, not present reality.

SPEAKER_01 16:25

This is incredibly heavy information. And just staring at these numbers without a defense strategy just creates paralysis. Since the corporate plans are already in motion, and since the CFOs have already filled out the spreadsheets, we need to extract an action plan from this data. Definitely. Looking at the synthesis of these three reports, how should an individual listener categorize their risk right now?

SPEAKER_00 16:48

The data reveals three distinct profiles in the modern workforce. Identifying which profile fits you is the first step in building a defense. Profile one is the high-risk aware worker.

SPEAKER_01 16:59

Okay.

SPEAKER_00 17:00

You work in an exposed industry like finance, programming, or media, and your daily output involves heavy digital processing. But you are aware of the threat. Awareness is your primary advantage because the tough displacement timeline is two to five years. You actually have a maneuverability window.

SPEAKER_01 17:15

Profile two is where the danger really lies, though.

SPEAKER_00 17:18

Profile two is the high-risk unaware worker. These are the writers, interface designers, and analysts who have maybe skimmed some vague news articles about AI, but they don't feel a sense of personal urgency.

SPEAKER_01 17:29

They suffer from the illusion of unique irreplaceability.

SPEAKER_00 17:31

Yes. And to them, the Duke CFO survey should be a blaring alarm. Your boss is in a closed-door planning meeting right now. The question is not if your department will see cuts, the question is when.

SPEAKER_01 17:44

Which brings us to profile three. If you're sitting in your car or at your desk right now thinking you fall into the lower risk complacent category because you work in a physically safe industry like healthcare or construction, you need to look really closely at your actual daily tasks.

SPEAKER_00 18:00

Absolutely.

SPEAKER_01 18:01

Are you delivering direct physical patient care, or is your job managing the administrative back office billing for that hospital? Do not confuse the safety of the industry with the safety of your specific desk.

SPEAKER_00 18:13

That is such a vital distinction. The receptionist managing the scheduling software at the plumbing company is entirely exposed, even if the plumber in the field is perfectly safe. Knowing your profile leads us directly into a four-step strategic framework that you can apply this week.

SPEAKER_01 18:27

What is step one?

SPEAKER_00 18:28

Step one is to quantify your specific risk. Do not guess. Go to the Tufts Index at digitalplanet.tufts.edu. Run your exact occupation and your specific city through their database. You need a hard percentage assigned to your livelihood, not an intuition.

SPEAKER_01 18:45

And step two connects back to the geography, right?

SPEAKER_00 18:47

Yes. Step two is to cross-reference your location with the wired belt data. If you are a data analyst living in a rural manufacturing heavy town, your displacement timeline might be slightly longer than an analyst living in Boston or Seattle.

SPEAKER_01 19:01

Because corporate adoption of AI moves at lightning speed in tech ups, the surrounding talent pool demands it. You have to know your geographic multiplier.

SPEAKER_00 19:08

Exactly.

SPEAKER_01 19:09

Step three is where we force the issue. Looking at the CFO data, the most logical defense mechanism for an individual right now is to force transparency. If 44% of C-suite executives are planning secret cuts this year, the move is to back them into a corner during your next one-on-one meeting.

SPEAKER_00 19:25

You have to ask a highly specific pointed question. Look your manager in the eye and ask, what is the company's explicit AI strategy for our department? And how does that affect the structural headcount of this team in the next 12 to 18 months?

SPEAKER_01 19:40

They'll probably dodge the question. They might use corporate speak about, you know, synergies and efficiency tools.

SPEAKER_00 19:46

But an evasion is data. If they cannot give you a clear structural answer about how AI will augment your specific role without eliminating it, that tells you everything you need to know about the conversations happening above their pay grade. Information is power.

SPEAKER_01 20:02

Which leads directly into step four. Build a lifeboat before the ship takes on water. You need to start building one new option in the next 30 days. We are not telling you to quit your job tomorrow, but you need to take a specialized course, secure an informational interview in a less automatable department, or develop a tangential side skill.

SPEAKER_00 20:20

And there is a piece of data from the ADP survey that completely validates this approach. ADP found that workers whose employers actively invest in their upskilling are 5.3 times more likely to feel secure in their jobs.

SPEAKER_01 20:31

5.3 times.

SPEAKER_00 20:33

Action cures anxiety. Engaging in deliberate skill building is the fastest psychological and practical way to move from the 78% who are terrified to the 22% who feel secure.

SPEAKER_01 20:45

The maps are officially drawn. The corporate plans have been made confidentially behind closed doors, but the numbers have leaked. You have the Doppler radar in your hands now? The only variable left in this equation is what you choose to do with this intelligence.

SPEAKER_00 20:59

Synthesizing all of this leaves us with a really provocative concept to consider. We are watching artificial intelligence rapidly automate the high-paying, white-collar knowledge work of the laptop class. Simultaneously, building the physical infrastructure to support that AI is triggering a massive surge in demand for the physical trades.

SPEAKER_01 21:18

We are looking at a total inversion of the American prestige economy. In a few years, as AI drives the market cost of generating code, writing reports, and analyzing data down to zero, the people we told to learn to code might end up wishing they had learned to weld. A commercial plumber or an electrician could soon hold more social capital, higher wages, and greater structural job security than a Silicon Valley software developer.

SPEAKER_00 21:42

It's a profound structural reversal.

SPEAKER_01 21:44

That is something you need to deeply consider as you plan your next career move. Share this deep dive with someone whose career could benefit from this intelligence because honestly, looking at the data, that is practically everyone. Thank you for listening. Be sure to check the show notes. For the links to the Tufts index and the ADP report, so you can find your specific numbers today. Run your occupation. Get the actual number. Do not operate on vibes when the data is available.

SPEAKER_00 22:10

Knowledge with that application is just trivia. Go find your number.

SPEAKER_01 22:13

See you on the next deep dive.

SPEAKER_00 22:21

Thanks for listening. Join us next time on Surviving AI.