AI Will Not Take Your Job, It Will Multiply Your To Do List

AI isn’t kicking down the office door with a pink slip. It’s buzzing your phone with 400 “helpful” drafts you now have to review by 5 p.m. That’s the strange truth behind today’s workplace anxiety: the apocalypse keeps getting predicted, but the lived reality feels like a rapidly expanding list of actions, approvals, and decisions. We dig into what’s actually happening with AI and jobs, using economic history, MIT style research framing, and a revealing March 2026 NBER working paper that surveys 750 corporate executives who control hiring and budgets. 

Along the way, we explain why cheaper intelligence doesn’t automatically buy us leisure. Jevons paradox shows how efficiency can increase total consumption, and we connect that idea to modern induced demand: when AI inference costs plunge, businesses unlock latent demand and suddenly “can afford” endless personalization, monitoring, market research, and scenario planning. Then we tackle the hard limit that keeps humans in the loop: Polanyi’s paradox. AI can devour explicit rules, but it struggles with tacit knowledge, common sense, and responsibility, which is why many of us become supervisors of brittle systems rather than beneficiaries of free time. 

The most disruptive shift may be hidden in plain sight: entry level roles built on routine tasks start to vanish, while senior workers become mega managers drowning in AI generated output. We end with what this means for your career and why relational labor, trust, negotiation, and judgment become more valuable as digital execution becomes table stakes. If this helped you see AI at work more clearly, subscribe, share it with a friend, and leave a review so more people can find the conversation.

Read companion articles and more tech analysis on Medium: https://medium.com/@allanandida

Leave your thoughts in the comments and subscribe for more tech updates and reviews.

Transcript

Allan 0:05

Picture the scene. You are uh huddled under your ergonomic standing desk.

Ida 0:11

Right.

Allan 0:12

The overhead lights are flickering. You are just waiting for the cold metallic clang of a chrome T eight hundred Terminator to kick down your office door and hand you a pink slip.

Ida 0:21

Oh yeah. The great sci-fi jobalypse we have all been bracing for.

Allan 0:25

Exactly. This is it. But then you know you peek out from the shadows, and the apocalypse looks, well, distinctly underwhelming.

Ida 0:33

Highly disappointing.

Allan 0:34

Right. Instead of a robot demanding your job, your phone just buzzes. It's a notification. Your newly implemented AI enterprise assistant has just seamlessly generated 400 highly targeted marketing leads, complete with personalized email drafts. Which sounds great until you realize Until you realize that now you, the human, have to personally review and approve every single one of them by 5 p.m.

Ida 0:56

It is the ultimate modern irony. I mean, we were promised an existential crisis, and what we actually got was a rapidly expanding list of daily action items.

Allan 1:04

So true. So welcome to this custom-tailored deep dive. Today our mission is to figure out what is actually happening under the hood with AI in our jobs.

Ida 1:13

Because there is a lot of noise out there.

Allan 1:15

So much noise. So we are analyzing a massive stack of recent sources for you. We've got MIT research, historical labor data going back centuries, and this incredibly revealing brand new March 2026 National Bureau of Economic Research working paper.

Ida 1:31

The NBER paper.

Allan 1:33

Yeah, the NBER one. They surveyed 750 corporate executives, you know, the people who actually control the budgets, to see what they are doing with headcount. We are looking for the truth beneath the panic.

Ida 1:44

Aaron Powell And the panic is very loud right now. You have this intense cultural anxiety. Just look at Ford's CEO Jim Farley recently predicting that AI is going to replace half of all white-collar workers in the U.S. Half. Right, half. When industry leaders throw around numbers like 50%, it's no wonder people are terrified.

Allan 2:02

Okay, let's unpack this. Because the data we are looking at drops a massive, entirely counterintuitive finding right at the top.

Ida 2:08

It really does.

Allan 2:09

The AI apocalypse. It has essentially been canceled, and it's been canceled due to an unprecedented increase in busy work. The machines aren't here to replace you. They are here to ensure you never, ever run out of things to manage.

Ida 2:23

It's true. The historical and current data consistently point to this one phenomenon. Cheaper, faster intelligence does not equal less work.

Allan 2:32

Right.

Ida 2:32

It just means we can finally afford to overcomplicate everything we do.

Allan 2:36

I was reading through these executive surveys, and people genuinely believe that doing their job ten times faster means they get to go home ten times earlier.

Ida 2:45

Oh, the optimism.

Allan 2:47

Right. Wait, it gets better. That's like buying a massive commercial-sized refrigerator and assuming you'll spend less time dealing with food. When really, you just buy way more groceries, start meal prepping for a month, and spend twice as much time cooking. Human nature just abhors a vacuum.

Ida 3:02

I actually love that refrigerator analogy because it mirrors a foundational rule of economics. To understand the mechanism behind this, like why efficiency doesn't lead to leisure, we actually have to look at the British coal industry in the 1860s.

Allan 3:15

Okay, Victorian coal mines. From generative AI to soot, walk me through it.

Ida 3:21

So in 1865, an economist named William Stanley Jevins noticed a trend that totally baffled people. Engineers had just developed steam engines that were dramatically more fuel efficient.

Allan 3:32

Okay.

Ida 3:32

So logically, the assumption was that the total consumption of coal would just plummet.

Allan 3:37

Right, because if the engine needs less coal to do the same work, we can just leave more coal on the ground. Makes perfect sense on paper.

Busy Work Cancels Job Doom

Ida 3:44

Makes perfect sense, but the exact opposite happened. Coal use skyrocketed.

Allan 3:49

Really?

Ida 3:49

Yeah. Because these new engines made coal-powered production so incredibly cheap, it suddenly became financially viable for entirely new industries to use steam power. Oh wow. They expanded the total market for coal. This is known as the Jevins paradox. When you increase the efficiency of a resource, you drop its price, which tends to increase, not decrease, the total consumption of that resource.

Allan 4:12

Aaron Powell Okay, so if we drag that into the modern knowledge economy, you're saying that making cognitive labor like synthesizing data or writing code more efficient doesn't shrink the demand for it.

Ida 4:22

Aaron Powell Not at all.

Allan 4:23

It blows the doors off the demand.

Ida 4:25

Aaron Powell It blows the doors completely off. And history is littered with examples of this. Take the late 1970s with the invention of electronic spreadsheets like VisiCalc.

Allan 4:35

Oh, sure. Before that, accountants were doing calculations by hand with erasers and physical ledgers.

Ida 4:40

Exactly. So people must have assumed software would wipe out the accounting profession overnight. Pundits were writing literal obituaries for accountants.

Allan 4:50

Let me guess. The Jevons paradox kicked in.

Ida 4:53

Big time. What actually happened was that financial modeling became so wildly cheap and easy to execute that businesses suddenly wanted vastly more of it.

Allan 5:02

Because they could afford to ask for it.

Ida 5:04

Right. Instead of asking an accountant to run one quarterly projection, a CEO could ask for 50 different what-if scenarios. The demand for financial modeling exploded and it ultimately quadrupled the number of accounting jobs.

Allan 5:17

Quadrupled.

Ida 5:18

We saw the exact same thing with ATMs. They made operating a bank branch so cheap that banks opened thousands of new branches, hiring more human tellers overall to staff them.

Allan 5:29

Or email. I mean, email was supposed to save us all the time we spent typing up physical memos and sending them through internal mail.

Ida 5:36

And how did that work out for us?

Allan 5:38

Horribly. The friction of communication dropped to zero, and now we are drowning in hundreds of emails a day.

Ida 5:45

Okay, but here's the thing isn't this just the concept of induced demand in traffic?

Allan 5:49

Oh, like highway expansion.

Ida 5:51

Yeah, exactly. Like when they spent over a billion dollars adding the carpool lanes to the 405 freeway here in LA, they thought it would solve congestion, but it just attracted all the drivers who previously avoided the highway.

Allan 6:03

And within months it was gridlocked again.

Ida 6:05

Right. Are we just building a massive 12-lane cognitive highway of AI only to sit in an intellectual traffic jam?

Allan 6:12

That is the perfect way to visualize it. We are building the 12-lane highway, and the toll to drive on it has essentially vanished.

Ida 6:18

The toll being the cost of computing.

Allan 6:20

Right. In the AI world, we talk about inference costs.

Ida 6:23

Yeah.

Allan 6:24

Training an AI model costs millions of dollars, but inference the cost of actually asking the AI a question and getting an answer has absolutely plummeted.

Jevons Paradox And More Work

Ida 6:32

How cheap are we talking? Hardware optimization and algorithmic efficiency have driven inference costs from dollars per query a couple of years ago to fractions of a cent today.

Allan 6:42

Which means companies are unlocking what economists call latent demand.

Ida 6:46

Yes. Tasks that a business could never justify paying a human$20 an hour to do are suddenly viable. Personalizing every single paragraph of a marketing email for 10,000 different clients.

Allan 6:58

Or generating hyperniche market research for a product that doesn't even exist yet.

Ida 7:02

Exactly. It's so cheap now, they just do it.

Allan 7:05

So they are overcomplicating everything because the execution is basically free.

Ida 7:08

Right. But here is the catch that breaks the system. Executing all that new overcomplicated work still requires human oversight, human direction, and human interpretation.

Allan 7:19

Which honestly brings up a glaring question for me.

Ida 7:21

What's that?

Allan 7:22

If we've built this 12-lane cognitive highway and the AI is driving the cars, why do we still need humans to hold the steering wheel at all? Like, if the AI can draft 400 marketing emails for fractions of a cent, why hasn't it just automated the approval and sending process end-to-end?

Ida 7:39

The barrier there is something called Polanyi's paradox. Coined by the philosopher Michael Polanyi in 1966, the paradox is simply this. We know more than we can tell.

Allan 7:50

We know more than we can tell. Give me a concrete example of that.

Ida 7:53

Think about the fundamental difference between explicit rules and tacit knowledge. Explicit rules are things you can write down step by step in a manual.

Allan 8:01

Like math calculations or basic data entry.

Ida 8:04

Or tax codes. AI devours explicit rules, but tacit knowledge involves intuition, common sense, and physical adaptability. It's cracking an egg into a bowl without shattering the yoke.

Allan 8:17

Or reading the tension in a boardroom and knowing when to crack a joke.

Ida 8:20

Exactly. We do these things effortlessly, but we don't explicitly know the exact neurobiology or physics of how we do that.

Allan 8:27

Oh, I see. It's like riding a bike. I can hop on a bike and ride it perfectly fine, but if you force me to write down the exact gyroscopic physics equations of how I maintain my balance while pedaling, I'd have no idea.

Ida 8:39

No clue.

Allan 8:40

The moment I try to codify it, I hit a tree.

Ida 8:42

That's it. And because human software engineers cannot explicitly write a step-by-step program to simulate a process, they only understand intuitively, automation fundamentally struggles with basic human adaptability.

Allan 8:54

So AI is basically brittle when it encounters the real world.

Ida 8:58

Notoriously brittle when it encounters tacit knowledge. There is this incredible experiment from Google back in 2012 that illustrates this hardware to intuition gap perfectly. It took a massive neural network of 16,000 processors just to successfully identify images of cats on YouTube.

Allan 9:18

Wait, 16,000 processors just to recognize a cat? Something a human toddler does with zero training data other than pointing at a pet.

Ida 9:25

Exactly the point. It is remarkably easy to make a machine pass a highly codified exam like the bar, but practically impossible to give it the common sense adaptability of a one-year-old child.

Allan 9:36

Okay, so how does this apply to the workplace?

Ida 9:38

It means automation rarely replaces the human entirely. Instead, companies frantically build intensely controlled, predictable environments like highly structured software pipelines so the machine doesn't get confused.

Allan 9:50

And then they use humans to babysit the edges.

Ida 9:52

Precisely. AI complements the non-routine work that requires judgment while entirely stripmining the explicit routine tasks.

Allan 10:00

Seriously. So we are essentially the uncodifiable common sense babysitters for these brilliant idiot savant machines.

Ida 10:09

That's a great way to put it.

Allan 10:11

But wait, if humans are the babysitters, which humans are we talking about? Because we still hear these massive numbers, millions of jobs at risk. Even if the aggregate number of jobs is stable, something massive has to be shifting underneath.

Ida 10:23

It is. And the data reveals a highly disruptive, frankly, darker tension inside the labor market. The total headcount might be stable, but the composition of those jobs is wildly shifting. Let's dig into that March 2026 NBER working paper with the 750 CFOs.

Allan 10:39

The folks holding the purse strings, what did they actually say?

Induced Demand Meets Cheap Inference

Ida 10:42

They reported practically no near-term aggregate employment decline due to AI. They noted maybe a 0.4% drop in large firms.

Allan 10:49

Which is basically statistical rounding error.

Ida 10:51

Exactly. And they actually saw modest employment gains in small firms because smaller companies are using AI to punch above their weight class and expand.

Allan 10:58

So the apocalypse we were all bracing for is basically canceled by paperwork. But if they aren't firing everyone, what is the dark tension you mentioned?

Ida 11:07

It's the evaporation of the entry level.

Allan 11:09

Oh.

Ida 11:10

Yeah. AI is brilliantly good at replacing explicit, codifiable routine tasks. And those just so happen to be the exact tasks that make up the bottom rung of the corporate ladder.

Allan 11:22

Walk me through what that actually looks like in practice.

Ida 11:24

Take a mid-sized law firm. Three years ago, a senior partner would ask three junior associates to spend their entire weekend reading and summarizing 400 contracts for a due diligence process.

Allan 11:36

The classic grunt work.

Ida 11:37

Right. Today, that same partner drags and drops those 400 PDFs into a secure, legal, large language model. They get the summaries back in three minutes, and they are 95% accurate.

Allan 11:49

So the partner is fine.

Ida 11:50

The firm doesn't fire the senior partner. In fact, the partner is more productive than ever, but the firm simply doesn't hire those three new graduates next semester.

Allan 11:57

Ah. So the corporate ladder still stands, but somebody quietly unscrewed the bottom rung.

Ida 12:03

And we are seeing the macro impact of this right now. Graduate unemployment hit a record 5.8% in 2025 in the US. That is a 30% jump since 2022. The junior analyst who spends all day clipping data into Excel, the trainee paralegal, the first year associate, those roles are vanishing as companies replace entry-level hiring with AI tools.

Allan 12:25

What does this say about us? I mean, we love cheap execution so much that we are willing to accidentally lobotomize our future leadership pipeline. Think about how learning actually works. If you cut all the junior roles today, if nobody's allowed to learn how to crawl by doing the grunt work, making the mistakes, and learning the context, who exactly replaces that senior law partner in 2036?

Ida 12:48

That is a structural nightmare currently keeping organizational psychologists awake at night. A senior engineer armed with an AI coding assistant is an unstoppable force multiplier.

Allan 12:57

Because they have 15 years of finely tuned tacit knowledge.

Ida 13:01

Right. They know instinctively when the AI is hallucinating or generating subpar inefficient code, but a newly graduated junior employee using that same tool.

Allan 13:09

They don't have the judgment yet.

Ida 13:11

No. They just accept the AI's plausible sounding answer, hit commit, and completely bypass the painful first principles reasoning required to actually master their profession.

Allan 13:22

They never build the tacit knowledge. They never actually learn how to balance on the bike. They just let the AI hold the handlebars until they crash.

Ida 13:30

Exactly.

Allan 13:31

And meanwhile, what is happening to the senior people? I imagine they aren't just kicking back and relaxing. The mega manager, I love it.

Ida 13:41

Because of that 12-lane cognitive highway of free execution, executives are seeing their workloads double. They are drowning in the sheer volume of AI-generated output they now have to oversee, tweak, and triage.

Allan 13:55

Because it's cheap to generate, but it still needs the common sense babysitter to sign off on it.

Ida 14:00

Exactly. A single AI agent monitoring customer behavior might generate 300 daily notifications, flag 50 potential churn risks, and draft 30 intervention emails.

Allan 14:11

And a human analyst still needs to read those, apply strategic judgment, and decide which ones to actually send.

Ida 14:17

AI doesn't eliminate decision making, it exponentially multiplies the number of decisions that need to be made by the humans left in the loop.

Allan 14:25

Okay, this brings up a massive contradiction that I think a lot of listeners are feeling right now. Junior employees can't get hired, and mega managers are absolutely drowning in notifications. So whatever happened to the utopian dream that AI would finally let us work less, I feel like we were promised a hammock and a four-day work week.

Polanyi Paradox And Tacit Judgment

Ida 14:44

Utopian optimism is incredibly persistent, despite all the evidence to the contrary. There was a recent Henley Business School survey where 57% of professionals boldly declared that AI integration will make the four-day work week a reality in the near future.

Allan 14:59

Let me guess the tech people were even more convinced.

Ida 15:02

Astonishingly so. Among IT professionals, 78% believe AI will bring about the four-day work week. Wow. And you see this translating into actual policy debates, too. Just recently, U.S. Senator Bernie Sanders cited potential AI job losses as a reason to propose a robot tax and push for a federally mandated 32-hour work week.

Allan 15:21

So policymakers are scrambling to address both the fear of mass unemployment and the hope for reduced hours.

Ida 15:27

Exactly. They're reacting to the anxiety.

Allan 15:29

Trying to legislate the hammock. But based on everything we've just discussed, the Jevons paradox, induced demand, mega managers drowning in triage, that four-day week is a complete mirage, isn't it?

Ida 15:40

History strongly suggests it is a mirage. Here is the reality check. As the cost of digital execution drops to absolute zero, the economic value of the human element skyrockets. We're seeing the economy pivot heavily toward what economists call relational labor.

Allan 15:58

Relational labor, you mean like empathy?

Ida 16:00

Empathy, trust, complex sales, negotiation, and genuine human connection. When every single competing company can instantly generate a flawless 50-page legal brief or a perfectly optimized marketing campaign, the output itself is no longer a competitive advantage.

Allan 16:16

It's table stakes.

Ida 16:17

Exactly. The premium currency of the new economy becomes the human who sits across the table from a client, looks them in the eye, reads the room using that tacit knowledge, and builds trust.

Allan 16:27

I was reading an essay by Matt Hopkins, and he points out a really cruel irony in all of this. The economy desperately needs these relational skills now more than ever, right? Right. But the younger cohort entering the workforce, the exact same people getting locked out of the entry-level jobs, are statistically spending significantly more time interacting with screens than with each other.

Ida 16:49

It's a fascinating and somewhat alarming collision of macro trends. We are racing toward an economy that desperately demands genuine human connection and nuanced judgment, populated by a younger workforce that is increasingly unequipped to provide it.

Allan 17:04

Because their entire digital lives have been optimized to minimize face-to-face friction. Exactly we've optimized the humanity right out of their daily interactions, and now we're realizing humanity is literally the only thing we can't automate. So how is the labor market bridging this gap? Are we just training people differently?

Ida 17:21

We are bridging the gap by creating the gloriously unnecessary, highly complex new categories of employment.

Allan 17:28

Oh, this is the best part.

Ida 17:29

To manage the chaos of the 12-lane highway, organizations are heavily hiring for titles that didn't even exist three years ago. Prompt engineers, AI ethics advisors, model fine-tuning specialists.

Allan 17:42

And my personal favorite from the research, AI orchestrators. I love that this exists, but also why? Are we really just inventing new job titles to herd unruly chatbots?

Ida 17:52

Pretty much.

Allan 17:53

It feels like we're replacing the people who used to write the code with people whose entire 40-hour work week is spent arguing with a machine about how it wrote the code.

Ida 18:03

It is simultaneously impressive and completely ridiculous, but yes, that is exactly what an AI orchestrator does. Because AI lacks tacit knowledge and common sense, we need humans to constantly steer, correct, and orchestrate the models.

Allan 18:15

It sounds exhausting.

Ida 18:17

Imagine your day-to-day job is managing a team of incredibly fast, highly confident, utterly literal interns who can read a million documents in a second, but who will also occasionally invent case law out of thin air because it sounded mathematically probable.

Allan 18:30

You'd spend your whole day double checking their work.

Ida 18:32

Which is exactly why the workload hasn't decreased. You are orchestrating chaos.

Allan 18:37

So synthesizing all this data, the CFO surveys, the Jevons paradox, the vanishing bottom rung, what does this mean for the listener navigating their career right now?

Ida 18:49

The main takeaway is actually profoundly reassuring, even if it's a bit exhausting. You are not facing a fiery apocalypse of unemployable masses.

Allan 18:57

You don't need to hide under your desk from a robot.

Ida 19:00

No, the data shows the jobs aren't disappearing. They're mutating.

Allan 19:03

We are just entering a hyper-productive, intensely overcomplicated era. The machines are doing the rote execution, which means the baseline of productivity has exploded.

Ida 19:13

Which forces you to step up into the roles that require deep judgment, strategic thinking, and emotional intelligence. They are basically forcing you to stop hiding behind spreadsheets and actually talk to your clients and make complex judgment calls.

Allan 19:27

Which honestly sounds a lot harder than just clipping data into Excel all day. We wanted to be saved from the grind of work, but AI just gave us a promotion to mega manager that we didn't ask for.

Ida 19:36

The ultimate irony of the AI revolution. Cheaper, faster intelligence just means the work shifts from quiet execution to loud oversight and relationship building.

Allan 19:47

So here is a final provocative thought for you to mull over as you go about your day. We've established that AI is driving the cost of flawless, instantaneous digital execution down to absolute zero.

Ida 19:59

Right.

Allan 19:59

When perfection is free, instant, and everywhere, what happens to the cultural value of human effort? Are we going to reach a point where corporate reports, customer service, or even art carry a premium luxury label that says 100% human-made?

Ida 20:15

Oh, that's interesting.

Data On Real Hiring

Allan 20:16

I'm talking about a label that proudly guarantees minor typos, slight inefficiencies, and a slower turnaround time, just like artisanal handcrafted coffee. Because if that chrome terminator ever does show up at the office, I have a feeling we're just going to make it sort through our inbox while we go grab a$14 human poured latte.

Ida 20:32

A wonderfully absurd but entirely plausible future.

Allan 20:36

Thanks for joining us on this deep dive. We'll catch you next time.