AI: What They're Not Really Saying

Sixteen developers. That's it.

Sixteen. Every one of them experienced.

Average five years working on their own open source projects.

These aren't tourists. These are people who know their code bases the

way you know the layout of your own kitchen in the dark.

Code bases averaging over a million lines of code.

Meter paid them a hundred and fifty dollars.

dollars an hour to do something very simple.

Do your real work. 246 real tasks.

Half of them use the AI tools.

The good ones. Cursor Pro running Claude 3.5 and 3.7 Sonnet.

The state of the art. Half of them,

no AI. And here's the thing. Before they started,

they were asked to predict how much faster will AI make you.

They said

24% faster. The economists who study this predicted

39% time savings. The machine learning experts predicted 38%.

Everyone agreed. The AI would make them faster.

Then they actually measured it.

Like actually measured it. Not a survey,

not a vibe, a gold standard randomized controlled trial.

We call that science.

The developers using AI took nineteen percent longer,

slower, not by a little, nineteen percent slower doing their own work with

the best tools money could buy at the time.

Then, after they finished, after they were objectively measurably

slower, they were asked again,

how did the AI do? And they said it made them twenty percent faster.

They were slower. They felt faster.

They could not tell the difference between the two.

Sixteen people who write software for a living on their own coat could

not perceive that a tool was slowing them down.

They were paid to notice. They were experts.

And their own senses lied to them in the same direction at the same time.

Now hold that thought, because right now across this country and across

the country I'm doing.

currently recording this, companies are laying off tens of thousands of people.

And the reason they give is that the AI makes everyone faster.

They can feel it. Stick with me.

This is The Overlap, the show about systems of power,

labor, and American injustice.

I'm Joshua, and I'm very glad you're here.

I am ~ actually recording this in Canada right now.

Don't worry, I haven't moved yet.

I don't think I'm going to, but I am recording this ~ remotely.

So the audio might sound a little bit different today.

The ~ content, however, is not going to be.

Because we still talk about systems of power and we talk about labor

and we talk about money and politics and how it relates to you.

Quick word for Will. He's still not with us.

We're sending our love and hope and all of those things to him out in the universe,

putting out some positive vibes.

I hope he's drinking a cup of coffee right now.

And listen to this. You earned it,

buddy. We love ya. So here's where we are.

The topic today is AI. Or more precisely,

it's the gap.

The gap between what we're told artificial intelligence is doing to

the economy and what we can actually measure it doing.

And I want to be clear from the top.

This is not an episode about whether technology is cool.

We all know it's cool. Some of it is really,

really impressive. This is an episode about proof.

It's about measurement. It's about the difference between a claim and a fact.

Because here's the thing that made me want to do this this week.

So this is July 2026. It's not one story.

It's never one story, right? It's a collision of two stories.

On one side, you've got the money.

For incredibly large companies,

Microsoft, Amazon, Google, Meta.

They are projected to spend $670 billion,

ba-ba-ba-ba-ba-billion, on AI infrastructure this year alone.

Six hundred and seventy billion dollars.

That's a sixty-two percent jump from last year.

And they have burned through their free cash flow.

They're actually issuing debt to keep building what they're building because they

think the value is going to eventually outweigh the detriment and what it's costing

them. Data center debt doubled to $182 billion last year.

On the other side of this, you have Goldman Sachs.

Not some, you know, skeptical blog.

it's not an Alex Jones scenario,

although the Onion has taken over ~ Alex Jones' old show,

and it's hilarious, you have to check it out.

So this is Goldman Sachs, right?

If you haven't heard of them, crawl out from under the rock you've been hiding

under. But their own senior economist said in March of this year,

we will not find a meaningful relationship between productivity and AI adoption

At the economy wide level. You didn't hear that wrong.

No meaningful relationship after all of that money.

Now the reason both the history and the moment matter here is that we have seen this

exact movie before, more than once.

And every single time the people telling you the revolution is here

and it's measurable, they were the people getting paid for you to believe it.

So we're gonna measure.

That's the whole game today. We're gonna do the one thing that the

the AI hype artists don't want you doing.

We're gonna ask quietly, patiently,

okay, show me where's the proof.

And if the claims were true, we'd have the proof by now.

But we don't. I want you to hold on to that phrase throughout this whole episode

because it's gonna keep coming back and it's gonna feel different every time.

If the claims were true, we'd have proof by now.

Not a forecast, not a survey, not a feeling.

Proof. So let's go back. Because to understand why this pattern keeps working,

you have to understand that it is a pattern,

and that has a shape. And the shape is older than most of the people building these

models. Nineteen eighty seven,

an economist named Robert Solaw,

Nobel laureate, MIT, one of the most respected people in this entire field,

wrote a short observation to

In the New York Review of Books.

And it becomes one of the most quoted lines in the history of economics.

He said, You can see the computer age everywhere,

except in productivity statistics.

Sit with that for a second. 1987,

I was three years old. By then,

American business had spent well over a decade pouring money into computers.

the nineteen seventies, the nineteen eighties,

offices are filling up with these machines that basically generate heat.

Everyone could see them, everyone believed they were changing everything.

The revolution was really visible on every desk.

But the productivity numbers flat.

Nothing. You could not find the computer revolution in the one place that

it was sold, that it was supposed to show up in the aggregate data.

The bottom line for the whole economy.

They actually called it the Sola Paradox,

and I want to be really careful here because the hype artists around

AI really love this story. They love it because they think it ends well for them.

They say see, the gains were real,

they just came late. Be patient,

the gains from AI are coming too.

And here's the thing that's half true.

And half that's true is the half that actually damns them.

Because yes, eventually the computer revolution did show up

in the productivity numbers in the late 1990s.

But I need you to listen to how it happened.

It took roughly a decade longer than the boosters of the eighties promised.

And it didn't happen because the machines got magic.

It happened because organizations spent years,

a decade, two decades restructuring how they actually

did work around the machine. New processes,

new training, new job descriptions,

new management layers built specifically to extract value from the

out of the technology that they had purchased.

The technology alone really did nothing.

The technology plus a decade of expensive,

painful, human organizational change actually did something.

That's the real solo paradox, right?

It's not be patient, the magic is coming,

it's the technology is the easy part and the productivity gain is the hard part

and it takes 10 years longer than anyone selling it.

is willing to admit. And importantly,

the companies that actually got the productivity gain weren't the ones that bought

the most machines. They were the ones that changed how people worked around

the machines. The hardware was necessary,

but not really sufficient. The human restructure,

that was the variable that really mattered,

and nobody selling hardware ever led with that,

because frankly that's expensive.

Now fast forward.

The 1990s, we probably remember this.

This is the dot-com boom. And ~ the claims.

The internet was going to permanently accelerate GDP growth.

Analysts predicted productivity gains of three to four percent every year forever.

Some of them said the business cycle itself was over.

Recessions, a thing of the past.

We had transcended economics. We were

We were post a post economic society.

There was a famous business week cover,

and there there were wired manuf manifesto,

serious, credentialed people with a lot of letters behind their name,

stood at podiums all over the world and said that the old rules no longer applied.

That the the internet had changed the underlying physics of growth.

The NASDAQ peaked in March of

two thousand

By October of two thousand two,

it had lost seventy eight percent of its value.

Five trillion dollars in market value were gone overnight.

Vaporized. Not over years, over months.

And the companies that disappeared weren't small bets.

Pets dot com WebVans

E toys.

These were businesses that had raised hundreds of millions of dollars

on the strength story, on the feeling that the internet changed everything.

So the old rules about revenue and profit and unit economics didn't really apply

to them anymore. And the productivity gains and the the the hypers promise

was you know, they came eventually,

sort of, but a decade later than it was actually advertised.

And nowhere near three to four percent forever.

The internet was real. It's still very real.

You're listening to this podcast over the internet.

The internet was transformative.

I'm not denying that. The internet is is ~ an infrastructure that's

now so incredibly foundational that we'd now kind of forget to notice it.

And the people who told you exactly how much and exactly when were wrong

by an order of magnitude and

And a decade. The technology is real.

It was real. But the timeline and the magnitude completely fabricated.

Confly stated, completely fabricated,

right? Same shape every time. So let me give you a third one because this one's

good, and I mean that in the the darkest dystopian way possible.

IBM's Watson. Between twenty eleven and twenty sixteen,

IBM

marketed Watson as this revolutionary AI that would transform medicine

and law and and finance. They spent billions of dollars promoting Watson health.

And here was the pitch. This thing can diagnose cancer better than your doctor,

better than a human oncologist,

the future of medicine running on a mainframe.

Indy Anderson Cancer Center, one of the best arguably cancer hospitals on the planet

Bought in all the way. They spent sixty two million dollars on

a Watson oncology project. Sixty two million to a research hospital.

That's not a rounding error. That's years of clinical trials.

That's research positions and equipment and opportunity costs that don't just come

back. They actually canceled it in twenty seventeen because it failed

to produce usable results.

sixty two million dollars for a system that couldn't do the one thing it

was sold to do. It's like if you spent sixty two million dollars on

a blender that didn't make smoothies.

And the Watson Health Division,

the flagship, you know, the future of medicine,

the thing that was going to cure cancer.

IBM actually quietly sold it off in 2022 for a fraction of what they they believed

it to be worth and hyped it to be worth.

There was not a press conference,

there was no, you know, Mia Culpa,

no accountability whatsoever. Just a quiet sale and a hope that you'd forgotten

the diagnosed cancer better than doctors' part.

The press releases from 2020.

13 or actually still on the internet.

~ I pulled a couple of them and you can read them too.

The gap between what was claimed and what was delivered is right there.

Archived and it's big. And it's kind of sad.

Did they know it might not work when they were selling it?

I I can't I mean I honestly I can't tell you what was in their hearts.

I can't tell you what was in their heads,

but I know the money moved before the proof did,

every single time.

The money moves first. The proof,

if it comes at all, comes years later.

And by then, the people who made the money are gone,

and the people who lost the sixty-two million dollars are writing it off as well,

it's a lesson learned. It's just make sure not to throw good money after bad.

One more, the self-driving car.

Between 2016 and 2018, Waymo,

Uber, Tesla, all of them repeat.

Repeatedly telling you full autonomy was two years away.

Two years. Always two years. Two weeks.

Travis Kalimek, the CEO of Uber,

predicted in twenty sixteen that self driving cars would replace every single Uber

driver within a decade. Elon Musk said in twenty nineteen that Tesla would have

a million robotaxis on the road by twenty twenty.

A million. By twenty twenty.

That decade is basically now. And as of twenty twenty six,

full level five autonomy, the car that drives everywhere,

anywhere, any conditions with no human,

it's not commercially available at scale.

It doesn't exist as a product you can buy.

Now Waymo operates you know in a handful of highly geofenced cities.

And it's pretty cool if you've ever had the opportunity to ride in one.

And

Outside of them, in the rain, in a construction zone,

in a small town in rural Oklahoma with no HD map data,

just doesn't work. Uber's self driving division actually got sold off

to Aurora in twenty twenty after one of their test vehicles killed

a woman in Arizona. You might remember that from the news.

The technology moved forward, but the promises from twenty sixteen,

they've yet to materialize.

Billions of dollars on the promise,

on two years away, and the investors who funded these promises just moved

on to the next story, and the drivers who were told their jobs would

be automated out of existence,

they're still driving. Now with the added anxiety of having been told

for a decade that they're just temporary,

they're a contract, they're 1099.

So here's the pattern. And let's name it clean before we walk into the present.

Because you're about to see it wearing a completely brand new suit.

The pattern is this a genuinely interesting technology shows up.

The people positioned to profit from it make enormous,

specific, confident claims about how much it will change the economy and how fast.

Those claims drive a flood of money.

And look, the money creates its own momentum.

You have to invest because everyone else is investing because

What if it's real?

And the actual measured results when they finally arrive are smaller

and slower and messier than anyone promised.

Computers in nineteen eighty seven,

the internet in two thousand, Watson in twenty seventeen,

self driving cars. We're still waiting on all of it.

Solo told us nineteen eighty seven,

and you can see it everywhere,

except the statistics. And now there's a machine that talks back.

And the claims are bigger than they have ever been.

So let me ask the question I I keep asking myself.

Why does this still work? Why,

after computers and the internet and Watson and the self-driving

car that's always two years away,

why does anyone still fall for this specific confident number thing?

And the oddest answer is because falling for it is extremely profitable

for the people telling you to fall for it.

That's it. It's the whole engine.

Let's follow the money because it's not subtle.

So Jensen Wong, the CEO of NVIDIA.

Now, if you don't know, if you're not involved in all of that,

you are hiding under a rock. NVIDIA makes the chips that everybody needs

to build these AI systems. And since ChatGPT launched,

NVIDIA's net income has grown roughly 20 times over.

That's a 20x return. Their market cap hit around $3 trillion.

And Jensen Wong is, without exaggeration,

the single biggest financial winner of the entire AI boom.

He is the arms dealer in a gold rush,

and he has been richly historically rewarded for it.

And here's the thing that he said that you should be sitting with

at night when you're waiting on your popcorn to finish.

Wong called the CEOs who blame AI for layoffs lazy.

Lazy. He said it doesn't even make sense that companies are already using

AI enough to be replacing people.

Play that back if you need to.

The man selling the shovels is telling you the gold isn't really there yet.

Meaning the person who profits the most from this AI race is saying the thing

the AI race is justified by just hasn't happened.

And he's still selling shovels as fast as he can make them.

He's even got some pre-sold for the future.

2026 to try to buy a chip is impossible because they're sold all the

way out until 2028.

That's why you see Anthropic and and Grok are renting out their servers

and and you know putting leases on Google Gemini to make it worse for everybody.

Is because they don't have the chips to to purchase.

They have to rent them from other people.

And look, Jensen Wong is not contradicting himself.

He is being perfectly consistent from the perspective of someone whose business

model is infrastructure, not the output.

He makes money whether or not

The productivity gain ever materializes.

Hy makes money on the belief.

And so he can afford to be honest about the belief's limits because

his balance sheet doesn't depend on the belief being true.

It depends on you buying the chips while you still believe.

And that's not a contradiction he's embarrassed by.

That's the business model. Sell the infrastructure for a revolution that

has yet to be proven. And if anyone get hurt gets hurt buying into it,

you just call them lazy. Now, the four big buyers are the hyperscalers,

right? Microsoft, Amazon, Google,

Meta. Six hundred and sixty-seven billion dollars in projected capex this year.

And I told you.

They've burned through their operating cash flow and started borrowing.

Data center debt doubled to $182 billion last year.

Why? Why keep spending when your own bank,

Goldman Sachs, tells you there's no measurable economy-wide return?

Goldman's James Cavello had the answer,

and it's pretty brutal. He diagnosed the whole thing as FOMO.

For those of you who don't keep up with internet slang,

FOMO means the fear of missing out.

He said, and I quote, FOMO has proven a stronger incentive than poor stock

performance. Think about what that means.

These are the four most sophisticated companies in the world.

They have armies of economists,

risk analysts, strategy teams,

and the thing driving six hundred and sixty-seven billion dollars in spending.

Is the same thing that makes a teenager buy a stock because their group chat won't

shut up about it. What if that's real?

What if the other guy figures it out first?

I can't be the one who missed it.

The CEO of a company with a trillion dollar market cap is,

at the moment of this decision,

operating on the same psychology as someone panic buying during

the GameStop GME rut.

That's not a strategy, right? That's panic.

Dressed up as a keynote. And then the third leg of the stool,

the consultants. Mmm, everybody's favorite.

McKenzie Global Institute. McKenzie projected that AI could contribute four point

four trillion dollars in productivity gains annually.

Trillion, that's with a T, you didn't hear me wrong.

Annually.

They anticipated three to four percentage points of additional GDP growth every year

from now to twenty forty. That report got cited in boardrooms everywhere.

It got cited by executives who had not read it to justify decisions that they

had already made. It was the permission slip.

four point four trillion dollars per year.

Hold that number.

Next to what an actual sober Nobel adjacent economist said.

Because in 2024, Darren Asamoglu,

MIT Nobel laureate, published a paper called The Simple Macroeconomics of AI.

In his estimate, AI would produce somewhere between one point one and

one point six percent of GDP growth over ten years.

Not per year, total.

over ten years. McKinsey said three to four points per year.

Asa Moblu said one to one and a half points over a decade.

That's not a disagreement or a rounding error.

That's two different worlds. McKinsey's number is roughly twenty

to thirty times bigger than than Asimoblu,

measured over the same horizon.

One of those two estimates is going to look very,

very wrong.

10 years. And I'll give you one guess which one was produced by

the firm getting paid to help companies make the spending decision.

And who pays McKinsey? The companies making the AI decisions.

That's who. Average monthly enterprise AI spend hit almost $63,000

in 2024, projected to rise to over $85,000 in 2025,

a 36% jump, driven in part by

By consulting recommendations.

Mackenzie helps produce the number that justifies the spending.

And then gets paid to help you do the spending,

and nobody, nobody, has a contract clause that says if the four point four trillion

doesn't materialize, we give your fee back.

The accountability runs in one direction.

That direction is the invoice.

Look, nobody ever goes back and checks Mackenzie's four point four trillion dollar

number against reality. You make the projection,

you cash the check, and by the time the number's proven wrong,

the engagement is over and there's a new number for a new technology.

The thing that holds this whole machine together makes me angry in

a very specific way. So in March of this year,

Goldman found that seventy seventy give or take percent of S

P five hundred management teams talked about AI on their earnings calls.

Seventy percent.

That means practically everyone's talking about it,

but only 10% quantified its impact on any specific use case.

And only 1%, 1% quantified its impact on earnings.

1% of people showed their work,

right? Showed the math. 70% of people are talking about it.

One percent can actually point to a number.

And the census data. So fewer than than twenty percent of US establishments

are even using AI for any business function at all.

Fewer than one in five businesses is using it.

But a major component of the biggest companies are are on stage talking about

it like it's already rebuilt the world,

like the restructuring is done,

like the productivity is already in the bag.

And that's the machine, right?

That's that's the thing.

It runs on talk, it runs on the number nobody checks,

and the FOMO nobody admits, and the earnings call where you have to say

the word AI enough times to keep the stock up,

whether or not you can actually point to a single dollar it's earned.

six hundred and sixty seven billion dollars and one percent can prove

it earned anything. If the claims were true,

we'd have proof by now.

We don't. So let's talk about Clairna because this is where it starts getting to,

you know, like real people and real things.

Because the AI hype artists are showing Clarna to everybody because they're

holding it up like a trophy. So Clairna's a Swedish fintech company,

financial technologies, buy now,

pay later. And they deployed an open AI powered customer service assistant,

Chatchi.

And then they went out and told the world about it.

And the numbers were incredible.

The AI handled 2.3 million conversations in its first month.

Clarina said it did the work of 853 full time human agents.

That means 853 people were essentially replaced by a chatbot.

They said the resolution time dropped from 11 minutes to under 2.

They said they saved sixty million dollars of twenty twenty five,

and they posted a hundred and fifty two percent increase in revenue per employee.

That was the case study globally.

All of these AI hype artists, they pointed at Clarina and said,

Hey look, here it is. The proof you keep asking for.

Here's a real company, real money,

real people replaced, real savings.

Stop being a skeptic. And look,

honestly, that's a really cool story.

If you only hear the first half.

Because the second half, which came in late 2025,

Clarin's own CEO, Sebastian I don't know,

Simiatikowski, admitted the company had,

in his word, over pivoted on AI.

Quality problems, the service got worse,

customers were having bad experiences.

Tickets that should have been closed in two minutes were getting mishandled

in ninety seconds instead. Is it faster?

Sure. Is it correct? No, it's not.

And they had to bring the human agents back to Clara.

Bring them back, like 853 people.

The trophy case study for AI replacing humans became a case study

in bringing the humans back because the AI started to falter.

So, because I mean this is kind of like the meter study all over again,

just at a scale of a whole company.

So Clarina felt the productivity gain.

They announced it, they put it in a press release.

They believed it's hard enough to restructure their entire customer service

operation around it. And then when the actual quality got measured

by their own customers having actual bad experiences,

reality decided to step in. This gain that they felt just evaporated,

and the humans had to come back.

The gap between what they believed and what was true cost real people their real

jobs. Temporarily, maybe, but you tell me.

You tell the customer service worker who got laid off in the trophy announcement

and quietly rehired in the walk back.

You tell them that that gap was harmless.

You tell them the months of unemployment and uncertainty while a company figured

out the AI wasn't actually doing the job that they built the AI for.

Tell the you tell them that it was a ~ it's a minor inconvenience and a story with

a happy ending. Fifty five percent of companies that cut jobs in AI

In 2024 and in 2025, they now regret it.

That's from HR Dive actually, 55% regret.

More than half. They fired real people based on what felt like a game,

and then they wish they hadn't.

And the the regret doesn't rehire anybody.

It's just an admission. It's a Mia culpa in the aggregate.

But the feeling wasn't proof.

Now let me tell you about a case where nobody gets to be rehired,

where the stakes aren't a customer service ticket.

The Centers for Medical Medicare and Medicaid Services,

CMS, in January of this year launched a program called Wiser.

And I I have to tell you the full name because the full name is kind

of a confession. Wasteful and inappropriate service reduction.

Reduction. That's the goal. Right there in the name.

Not better medicine, not more accurate decisions,

reduction. The metric baked into the program's identity is denying more things,

not getting more things right.

Wiser uses AI to evaluate and effectively to automate the denial

of requests for medical care for people on Medicaid.

Poor people, sick people, people asking for care they and their doctors believe they

need.

And what should make your skin crawl is that the program was deployed with

no public transparency about how the AI made its decisions.

No published error rate, no independent audit before launch,

no clear, accessible path for a patient to appeal an algorithmic denial.

Nothing. A machine is deciding whether a poor sick person gets care,

and the public, the people who essentially pay for it.

Is not allowed to know how well the machine works.

We don't know how often it's wrong.

We don't know what it optimizes for.

We don't know whether it was validated against actual patient outcomes

or just against prior authorization approval rates.

We know the name of a program,

and we know it denies things, that's about it.

Did you see it? Do you see how the hype connects to this?

The whole reason.

A government agency can deploy a black box to deny medical care,

is that we've spent three years being told these systems work,

that they're smart, that they're better than the humans.

Nobody demanded the error rate because the culture had already assumed

the machine was competent. The hype did that.

The hype built the permission structure for a machine to deny your grandmother's

care without ever having to prove that it can tell a valid claim from an invalid.

And the hypers primed the audience.

And CMS walked through the door and the hypers opened it.

The Electronic Frontier Foundation,

the EFF. If you haven't heard of these guys,

you should really check them out.

They do some interesting work.

They filed a FOIA request, Freedom of Information Act lawsuit,

against CMS to try to get the records to find out how this thing actually decides.

They just wanted to see the algorithm,

they wanted to see the training data,

the validation results, the error rates,

the basic information you would want before trusting a system with that

big and that consequential of a decision.

And as of this month, July of twenty twenty six,

the lawsuit is still pending, still.

They're still fighting just to see the numbers and how it was trained

and whether there are even numbers at all.

And that's the through line made physical.

The one percent problem. Remember,

only one percent of companies can actually quantify AI's impact on earnings?

Well, CMS can't. Well, maybe they won't.

Quantify Weiser's impact on patients.

Same darkness. Same refusal to measure.

Except now the thing being measured isn't a stock price.

It's whether a human being gets the care that keeps them alive.

Remember the 16 developers from the cold open?

Experts, paid to notice, who couldn't feel that a tool was slowing them down.

The perception gap wasn't malice.

It was just human. People feel the tool working,

they feel the responses generating,

they feel productive. And the gap between the feeling and the measurement was 19%.

Now imagine.

That same perception gap, that same confident wrongness sitting inside

a government system that denies medical care to people who could

not otherwise afford medical care.

Nobody at CMS is going home feeling like they made wrong decisions.

The machine says no, it feels efficient,

it feels like fiscal responsibility,

which feels like progress. And there's no meter study for.

People wiser denies. There's no gold standard trial,

no, you know, $150 an hour to sit down and measure whether the machine is right.

There's just a denial and a sick person and a lawsuit still pending.

If the claims were true, we'd have proof by now.

They're denying care with a machine that they won't let anybody measure.

So let me pull this all the way back and I will tell you what I think

of this and what it actually is.

Because I actually don't think it's about AI.

AI is just the current, you know,

costume, it's the flavor of the week.

The pattern is the perception reality gap.

And it runs in the exact same direction every single time.

What people believe about the productivity gain is always,

always bigger than what you can measure.

And the industry, this is the the the

important part. The industry systematically avoids funding

the measurement that would close the gap.

Think about the the meter study again.

It's the first gold standard randomized controlled trial of AI

on real developer work. The first one.

And it came out in July of twenty twenty five.

That's why if you're following all the the you know foundational models,

it used models that are now deprecated and older.

But

ChatGPT launched in November of 2022.

Before that, we didn't have anything like this.

That's more than two and a half years of the biggest technology story on Earth.

Hundreds of billions of dollars.

Entire companies restructured before anyone ran

one rigorously controlled experiment on whether the flagship use case actually makes

people faster. Two and a half years.

And the gap wasn't filled by the companies building the tools.

GitHub's own study, the vendor study,

claimed a 55% productivity boost.

55%. Now that was based on self-reported feelings

at narrow synthetic tasks that looked nothing like real development

and production work. That number was cited everywhere,

like in earnings calls and board presentations and layoff announcements.

For years it was the number.

The vendor's feeling based number,

right? The one before the independent trial existed.

And when they finally ran the independent trial,

the answer was 19% slower. Why do you think it took so long to measure that?

Why do you think the measurement that existed was a vendor survey

and not a controlled trial? Because the measurement was the enemy in this scenario.

The gap is the product.

As long as everyone's operating on vibes,

the perception always favors the sale.

The moment you measure, the moment you run the controlled trial,

the moment you you know FOIA the algorithm,

that gap starts to close. And the closing always goes the same direction,

down, toward reality, toward smaller and slower and messier than promised.

Solov told us in 1987,

everywhere except the statistics.

And here's the human cost. And this is where it stops being

an interesting economic story and starts being an injustice,

which is what this show tries to point out for everyone.

The Harvard Business Review surveyed over a thousand global executives,

and the finding was devastating in how plainly they said it.

Companies are cutting jobs because of AI's anticipated future potential.

Not because AI has demonstrated it can do the displaced workers' job anticipated.

Future potential. These are not words that describe a thing that's been proven.

Those are words that describe a bet.

And the people losing their jobs aren't losing them because the machine

has proven it can do the work.

They're losing their jobs because a board decided the machine might do it someday.

And someday is cheap, and people are expensive.

Nearly a hundred thousand jobs since late twenty twenty three were blamed on AI.

fifty five thousand in twenty twenty five alone.

By May of this year, over eighty seven thousand layoffs blamed on AI just this year.

Tech sector unemployment climbed to five point eight percent.

That's the highest tech unemployment has been since the dot com bust.

The last time we did exactly this thing.

The last time the feeling outran the reality and regular people paid

the price while the infrastructure sellers cashed out.

An anthropic. Anthropic is an AI company doing the research.

They found that job finding rates for workers aged 22 to

25 entering AI exposed jobs fell

14% since ChatGPT launched.

Young people trying to start a career locked out of the first rung of the ladder.

Not because a machine is doing the job really well.

But because a hiring manager has a feeling that a machine might do

it really well eventually, the cost of the hype is landing on the on

the people who can least afford to absorb it.

Not the executives deciding to cut,

not the investors funding the infrastructure,

the twenty three year old who can't get their first interview.

That's the injustice.

The gains are theoretical, the layoffs they're real.

The $4.4 trillion is a McKinsey slide in their latest,

you know, keynote. The 87,000 people are human beings with rent

and student loans and families who were counting on a job because they were promised

that if they went into this field and got this degree and went into this debt,

that there would be something they're waiting for them.

We are transferring wealth and security away from workers toward NVIDIA

and hyperscalers and consultants,

and we're doing it on the strength of a of a vibe.

A vibe that sixteen experts couldn't even trust with their own work.

A feeling Clarina believed until their own customers proved it wrong.

A feeling CMS won't even let us look at while it denies people health care.

The perception reality gap isn't a quirk.

It's not an accident. It's a mechanism for moving money.

And money always moves in a very specific direction,

we've said that before, away from people who work.

Word people who own. And the people it moves money away from,

they don't get to file a lawsuit and wait.

They don't get a FOIA request,

they just lose the job. And then they watch the earnings call where the

CEO says the word AI 14 times and the stock goes up.

So here's what I want you to watch more closely.

Not the claims, the measurement.

Or, in this case, the absence of it.

Every time you hear a number about what AI is doing to productivity or jobs

or the economy, I want you to ask one question.

Who measured this? And can I see it now or am I going to see it later?

If the answer is a company survey or a self reported feeling

or a McKenzie projection about twenty forty,

that's not proof. That's marketing.

What I want you to carry out here is this too.

Seventy percent of the biggest companies in America talk about AI on earnings costs.

One percent can point to what it did to their earnings.

Seventy talking, one numbing. Hold those numbers next to each other for

the rest of this year, and you'll see the whole thing pretty clearly.

And then Jensen Wong, the man who's made more money off of this boom than anyone

alive. Even, yes, your beloved and beloved to be hated Elon Musk,

whose net income grew 20 times over selling picks and shovels.

He's the one telling you the gold isn't there.

He called the CEOs blaming AI for layoffs lazy.

He said it doesn't make sense that AI is replacing people already.

The biggest beneficiary of the hype is on record saying the hype's central claim

is false. And he said it while selling more shovels.

That's not a slip. That's just like the entire logic naked.

You can profit enormously from a revolution you don't even believe has come,

as long as everyone else is too afraid to miss out to stop and measure.

So two things.

radical over a drink things. One,

if you work somewhere that's making AI decisions,

ask the right questions for measurement.

Not like a vendor deck, the internal number.

What did we measure and how? And then watch what happens.

In a lot of places, a question is going to be the most disruptive thing that

you do because making the Emperor's tailor put something in writing,

the paper trail matters, you know,

later. And two, if you're not if you're hiring,

Or you know someone else who is,

go and hire the twenty two to twenty five year olds getting locked

out of this economy right now.

The ones down fourteen percent on job finding because a machine might someday maybe

be able to do their job adequately.

That's the most ~ underpriced talent in the market.

So everybody's too spooked by a feeling to grab it.

Be the person who measures instead of a person who feels.

If the claims were true, we'd have proof by now we don't.

So stop accepting that vibe and start asking for the number.

This has been the overlap. You can find us at FOF.foundation.

We are available on Apple Podcasts and Spotify,

and you know, wherever you like to get your podcasts.

Will, I hope you're enjoying yourself.

To the rest of you, go ask somebody to pull out a ruler and measure something.

Take care of one another. Bye.