Matt Shumer's viral essay "Something Big Is Happening" hit 80 million views on X. The thesis: AI is about to change everything.
I think the tense is wrong. Something big is not coming. Something big is already here.
The Papers of Record Are Noticing
The Atlantic. Business Insider. The Wall Street Journal. These are not fringe tech blogs chasing engagement. These are legacy institutions with editorial boards and lawyers. And they are all running the same story now, from different angles, arriving at the same conclusion: white collar work is being restructured by AI, and it is happening faster than the frameworks can track.
When one outlet publishes a think piece, it is a take. When all three converge within weeks of each other, it is a signal. The Overton window has moved. The conversation is no longer "will AI affect knowledge work?" It is "how much has it already?"
The Question I Cannot Answer
Here is where I lose the thread of my own certainty.
I have spent months writing about amplification. About leverage. About how the right operator with the right tools can produce at the scale of a team. I believe that. I have lived it. My own output with AI is proof of concept.
But when I read the aggregate coverage — when I see the data on hiring freezes in legal, finance, consulting, content, and software — I cannot cleanly separate the amplification story from the replacement story. Because both are true simultaneously, and the ratio between them is shifting in real time.
Is this a mass replacement event or an amplification event for productivity?
I do not know. And I am suspicious of anyone who claims they do.
The Honest Math
Here is what I can see clearly.
If one person with AI can do the work of five, that is amplification for the one and replacement for the four. The narrative you tell depends entirely on which seat you are sitting in.
The CEO sees a productivity revolution. The mid-level analyst who just got a thank-you email and a severance package sees something else entirely. Both are describing the same event accurately. That is what makes this so hard to parse.
Previous technology shifts had longer runways. The internet took a decade to restructure retail. Cloud computing took years to reorganize IT departments. This is moving on a timeline measured in quarters, not decades. The adjustment period that normally gives workers time to retrain, pivot, and adapt is being compressed into something that feels less like a transition and more like a cliff.
My Concerns
I am not a doomer. I build with these tools every day. But I have three concerns that I cannot logic my way out of.
First: the amplification benefits are not distributing evenly. The people who gain the most from AI leverage are people who already think in systems — architects, strategists, founders, senior operators. The people most exposed to displacement are the ones who were executing within systems designed by others. The same hierarchy that existed before AI is being accelerated by it, not flattened.
Second: the speed is outpacing the social infrastructure. Retraining programs, unemployment systems, career transition support — none of it was built for a disruption cycle this fast. We are asking people to "learn to use AI" as if that is a weekend course and not a fundamental rewiring of how they relate to their own professional identity.
Third: the public conversation is still stuck in binary. Either AI is a tool and everything is fine, or AI is a threat and we should panic. The reality is that it is both, for different people, at different levels of the stack, at different speeds. And the inability to hold that complexity in public discourse means policy will lag reality by years.
The Naysayer Has a Point
There is a reasonable pushback to all of this, and I want to give it its due because it is not wrong.
AI is right about 90% of the time. The other 10% it confidently fabricates something that looks indistinguishable from truth. In high-stakes domains — law, medicine, finance, infrastructure — that 10% is not a rounding error. It is a liability. And the naysayer's argument is simple: you cannot hand the keys to a system that is wrong one out of ten times and call it a revolution.
They point to entrenched systems. Regulations. Compliance frameworks. Professional licensing. Industries where the cost of a wrong answer is not an embarrassing email but a lawsuit, a misdiagnosis, a bridge that fails. These systems move slowly for reasons. The friction is not bureaucratic laziness — it is a load-bearing wall.
And they are right. About all of it.
But here is where the framing breaks down. The naysayer is answering a question nobody is actually asking. The question is not "will AI replace the lawyer?" It is "will AI replace the four paralegals, three research associates, and two junior attorneys who support the lawyer?"
We will still need humans. We will need editors who catch the 10% the model gets wrong. We will need safety engineers who validate what the system produces. We will need lawyers who understand liability, doctors who make judgment calls, architects who sign off on plans. The human in the loop is not going away.
What is going away is the infrastructure of humans around that human.
A senior attorney who used to need a team of eight to prepare for trial can now do the research, document review, and brief drafting with AI and a single sharp associate. The attorney is more productive than ever. The seven people who used to support that workflow are not needed at the same scale.
That is not replacement of the expert. It is the hollowing out of the support structure beneath the expert. And that is where the job losses actually live — not at the top of the pyramid, but in the middle and at the base. The people who did the work that AI now does at 90% accuracy with a human checking the last 10%.
The naysayer looks at the 10% error rate and says "see, you still need us." And they are correct — you still need some of them. The question that should keep everyone up at night is: how many?
What I Know
I know the amplification is real because I am living it. I know the displacement is real because the data is showing it. I know these two truths are going to coexist uncomfortably for a long time, and I know that the people who insist on collapsing it into a clean narrative — either utopian or apocalyptic — are not helping.
I also have canaries. HawaiiGuide — a travel site I have run for over twenty years — is down 57% in traffic. Our gardening site is down about 40%. Both had been growing steadily until around 2024, when things shifted and then accelerated with AI Overviews. Not because the content got worse. Not because competitors outranked us. Google Search Console says our ranking positions are stronger than ever. The clicks just stopped arriving. If that is not a harbinger, I do not know what is.
The media convergence is the tell. When The Atlantic, Business Insider, and the Wall Street Journal are all saying the same thing at the same time, the signal is no longer debatable. The only question is what comes next.
I will keep building. I will keep writing. And I will keep being honest when I do not have the answer, because right now, on this one, I do not.
The Protocol: Something big is not coming. Something big is here. The honest position is that it is simultaneously the greatest leverage event and the greatest displacement risk of our professional lifetimes — and anyone who tells you it is cleanly one or the other is selling something.
