YOU DON’T GET TO BLAME AI FOR THE WORLD YOU ALREADY BROKE
The machine arrived after the damage. Bring the chronology.
No.
You’re not doing this.
You’re not going to spend decades building a society ordinary people increasingly cannot afford to participate in, introduce artificial intelligence near the end of the timeline, and then point at the machine like it walked into a perfectly functioning civilization and broke everything.
The dates don’t work.
AI didn’t make your houses unaffordable.
AI didn’t spend decades turning shelter into an appreciating financial asset that increasingly outran the incomes of the people expected to live in it.
AI didn’t create the mortgage bubble.
AI didn’t create the financial crisis.
AI didn’t teach an economy to confuse asset appreciation with productive progress.
AI didn’t create medical bills.
AI didn’t create student debt.
AI didn’t create institutional distrust.
AI didn’t create financialization.
AI didn’t create the increasingly absurd distance between what civilization is technologically capable of producing and what an ordinary working person can actually afford.
You did all of that while the machine was still science fiction.
And now the machine is here.
How convenient.
BRING THE OLD NUMBERS
This is the part that makes the story so ridiculous.
We don’t have to speculate.
You kept records.
The Federal Reserve’s own historical material shows mortgage debt relative to disposable income sitting around 40% during much of 1970–1985, rising toward 60% around 1990, and reaching roughly 100% by 2005. (Federal Reserve)
That wasn’t ChatGPT.
By the eve of the financial crisis, the Fed’s household debt-service series shows total household debt payments consuming roughly 15.9% of disposable personal income in late 2007, with mortgage debt service alone near 9%. (Federal Reserve)
Still no superintelligence.
Then look at housing itself.
Federal Reserve analysis using the Survey of Consumer Finances found the median home worth more than 4.6 times median family income in 2022, exceeding even the 2007 peak of 4.2. The Fed explicitly notes that rising home values mean declining affordability for would-be buyers. (Federal Reserve)
The machine didn’t do that either.
The warning light wasn’t blinking.
The engine had already been smoking for decades.
AND PEOPLE ALREADY KNOW SOMETHING IS WRONG
Here’s where the new narrative runs into another problem.
You need people to trust the narrator.
They don’t.
Gallup reported in July 2026 that average confidence across 14 major American institutions was 27%, only one percentage point above its historical low. (Gallup.com)
Twenty-seven percent.
You don’t have a messaging problem.
You have a credibility problem.
And look at what ordinary people say they’re worried about right now.
Pew’s July 2026 survey found only 24% of Americans describing economic conditions as excellent or good. Sixty-nine percent were very concerned about healthcare costs, 66% about food and consumer goods, and 64% about housing. Concern about AI’s role in the economy was 49%. (Pew Research Center)
Read that again.
The people are telling you what hurts.
Healthcare.
Food.
Housing.
Energy.
Actual things.
Things attached to bodies.
Things attached to houses.
Things attached to refrigerators.
Things attached to electrical outlets.
Things you cannot press-release into existence.
And somehow the prestigious conversation keeps floating upward into mythology.
HERE COMES THE ADULT SANTA CLAUS
Now, in September 2026, some of the most powerful AI institutions are publicly discussing recursive self-improvement, catastrophic capabilities, mandatory national safety requirements, and circumstances under which AI development should slow or stop.
OpenAI’s own September 9 policy statement says fully autonomous recursive self-improvement is not happening today, while arguing that governments should prepare for that possibility and establish mandatory capability-based safety regulation. (OpenAI)
Fine.
Study it.
Test it.
Secure dangerous systems.
Do legitimate engineering.
But don’t you dare erase the chronology.
Because there is an enormous difference between:
AI may create new problems
and
AI explains why your society already looks like this.
It doesn’t.
The machine arrived at the crime scene late.
YOU WERE ALREADY HERE
That’s the sentence that destroys the escape hatch.
We were already here.
The houses were already expensive.
The debt was already accumulated.
The institutional confidence was already collapsing.
The financial abstractions were already enormous.
The provenance problem already existed.
The databases were already mutable.
The incentives were already distorted.
People were already wondering why extraordinary increases in technological capability weren’t translating into corresponding increases in ordinary material security.
Then AI showed up.
And suddenly we’re supposed to discuss whether the machine is going to destabilize society?
Brother.
Look around.
THIS ISN’T AN AI STORY
It’s an allocation story.
For decades, society has been sending signals about what it rewards.
Attention can be monetized immediately.
Speculation can be monetized.
Intermediation can be monetized.
Asset appreciation can be monetized.
Narrative can be monetized.
Status can be monetized.
Existing scarcity can be monetized.
But somebody trying to produce a genuinely new capability can spend years being asked:
Who funded you?
Who validated you?
Which institution recognizes you?
How large is your market?
Who else believes you?
Where are your credentials?
Where is your traction?
Where is your permission slip?
That’s backwards.
The civilization eventually becomes extraordinarily sophisticated at pricing representations of value while becoming increasingly confused about the production of value itself.
And then everybody acts surprised when the scoreboard starts looking insane.
ENTERTAINMENT ENTERTAINS. INVENTION SUSTAINS.
That isn’t morality.
It’s mechanism.
Entertainment has value.
Art has value.
Sex has value.
Sports have value.
Stories have value.
None of that is the argument.
The argument is that civilizations require certain activities merely to remain civilizations.
Someone has to produce energy.
Someone has to move water.
Someone has to grow food.
Someone has to build housing.
Someone has to maintain infrastructure.
Someone has to discover medicines.
Someone has to manufacture machines.
Someone has to create new productive capabilities.
Someone has to preserve knowledge accurately enough that the next person doesn’t have to rediscover everything from scratch.
Reality has dependencies.
Your culture can assign status however it wants.
Physics doesn’t care.
AND THAT IS WHY THE RECORD MATTERS
Here’s the deeper failure.
We don’t even maintain a clean, universally verifiable genealogy of creation.
Who actually made the thing?
When?
What exactly existed?
What was its predecessor?
What changed?
Who possessed authority over it?
What evidence survives?
Can the artifact demonstrate its own history?
Or do we have to ask a platform, institution, database, publication, repository, corporation, credentialing body, or historian what happened?
That’s not a trivial archival problem anymore.
Because now you’re putting machines on top of those records.
And machines can operate faster than bureaucracies ever could.
So an institutional representation becomes machine context.
The machine acts on it.
The action generates another record.
Another machine consumes that record.
The representation becomes an input into reality.
You have automated the feedback loop.
Congratulations.
You didn’t create artificial intelligence in a vacuum.
You connected it to civilization’s accumulated assumptions.
AND NOW YOU WANT TO TELL ME THE MACHINE IS THE SCARY PART?
The machine needs electricity.
The machine needs chips.
The machine needs cooling.
The machine needs datacenters.
The machine needs networks.
The machine needs humans to build those datacenters.
The machine needs humans to maintain the grid.
The machine needs somebody paying the bill.
Turn the infrastructure off and watch how quickly the digital deity remembers that it has a power cord.
Yet we’re supposed to talk like we’ve summoned an immortal spirit.
It’s adult Santa Claus.
A machine with dependencies gets narrated as an independent metaphysical force while the extremely non-metaphysical civilization underneath it struggles with housing, healthcare, food, energy and trust.
Maybe inspect the floor before giving another conference about the ceiling.
AND HERE’S THE PART THAT REALLY HURTS
You can’t blame the machine for the timeline preceding the machine.
That’s what chronology is for.
If employment deteriorates after AI adoption, measure the change.
If wages change, measure them.
If productivity changes, measure it.
If inequality changes, measure it.
If information quality changes, measure it.
If AI causes harm, attribute the harm to AI.
Absolutely.
But establish the predecessor state first.
Because you don’t get to inherit a fifty-year-old problem, attach a neural network to it, and rename the entire lineage:
AI CRISIS.
No.
Bring the predecessor.
Bring the dates.
Bring the numbers.
Bring the state before deployment.
Bring the transition.
Bring the successor.
Then we’ll know what the machine actually did.
THE PUBLIC DOESN’T NEED YOUR STORY
That’s perhaps the greatest miscalculation.
People live inside the measurement.
A person doesn’t need an economist to explain that the house they cannot afford is unaffordable.
They don’t need a think tank to explain their grocery receipt.
They don’t need a futurist to explain their medical bill.
They don’t need an AI safety conference to explain their electricity bill.
They don’t need a prestigious institution to tell them whether they trust prestigious institutions.
They’re the ones experiencing the output.
That’s why institutional confidence at 27% matters. (Gallup.com)
You can only maintain a contradiction between representation and lived reality for so long before people stop arguing with the representation.
Eventually they just stop believing you.
That isn’t ignorance.
That’s settlement.
THE SEVERE OUTCOME IS NOT COMING.
YOU’RE STANDING IN IT.
That’s what everybody keeps missing.
They’re waiting for some cinematic collapse so they can finally declare that something went wrong.
Look backward.
Measure what ordinary productive participation purchased.
Measure what it purchases now.
Measure housing against income.
Measure debt.
Measure family formation.
Measure infrastructure.
Measure institutional confidence.
Measure the accumulated technological capability of civilization.
Then ask the question nobody wants to ask:
Given everything humanity learned and built over the last seventy years, should ordinary life have become this difficult to secure?
Don’t compare 1958’s television to an iPhone.
That’s childish.
Compare productive capability to human outcome.
We can manufacture semiconductors with features measured in nanometers.
We can communicate across the planet instantaneously.
We can automate intellectual labor.
We can manipulate genomes.
We can land reusable rockets.
We can operate machines containing hundreds of billions of parameters.
And somehow shelter remains an existential financial problem for millions of people.
That’s not technological scarcity.
That’s an allocation question.
SO NO.
You don’t get to introduce AI at the end of the movie and make it the villain of Act One.
The receipts survived.
The mortgage data survived.
The housing data survived.
The debt survived.
The institutional polling survived.
The chronology survived.
And now, ironically, you’ve handed ordinary people machines capable of searching all of it.
So go ahead.
Tell everyone how powerful AI is.
Tell them it can reason across enormous quantities of information.
Tell them it can search history.
Tell them it can analyze datasets.
Tell them it can connect patterns humans overlook.
Then watch somebody type the most dangerous prompt imaginable:
“Show me the numbers from before AI.”
Oops.
The machine didn’t create the world it woke up inside.
You did.
And the chronology isn’t going to take the fall for you.
