AI Has Reached the Asmongold Layer
- Rich Washburn
- 1 day ago
- 9 min read
Somewhere between the research lab and the corporate boardroom, artificial intelligence entered a room with an unmade bed, a long-haired man and absolutely no remaining tolerance for bullshit.
I have to admit something. Until very recently, I did not really know much about Asmongold. I knew the face. You know the face. The long hair. The thousand-yard stare. The general appearance of a man who has been awake since the Bush administration and has developed a deep personal distrust of shampoo, corporate communications and probably several forms of furniture.
I had seen him while scrolling. I knew he was internet-famous. I knew he reacted to things. I knew millions of people watched him. I had just never clicked. Then I clicked.
That was a mistake. Not because he is bad. Because the guy is absolutely hilarious, and now I understand how people lose three hours watching him interrupt a 22-minute documentary.
Asmongold does not merely react to a video. He attacks it from the side. He pauses every 17 seconds, says the exact thing nobody in polite company is supposed to say, occasionally wanders into a completely unrelated story about a discontinued consumer electronic from 2003, then somehow circles back and delivers a sharper analysis than the people who produced the original video. It is chaos. Beautiful, greasy, strangely intelligent chaos. And after watching him tear through a ColdFusion piece about artificial intelligence replacing workers, hallucinating inside businesses and generally failing to live up to some of the more breathless promises being made about it, I came away with two conclusions.
First, Asmongold is a riot. Second, I think the guy may be a freaking genius. Not in the traditional tweed-jacket, leather-elbow-patch, "let us examine the epistemological framework" sense. I mean something more useful. He has a remarkably good bullshit detector. And right now, artificial intelligence desperately needs people with good bullshit detectors.
He Is Not Anti-AI
That is the first thing worth understanding. Asmongold is not sitting there yelling that AI is fake, useless or some elaborate Silicon Valley scam. Quite the opposite. He repeatedly acknowledges that AI is real, powerful and likely to transform business. He sees it as a force multiplier. One person may be able to do the work of several people. Repetitive work will be automated. Entire industries will change. Companies that refuse to use it may eventually be outcompeted by companies that do.
That is not skepticism. That is realism. His objection is to the enormous group of executives, consultants and newly self-declared visionaries who have taken the inevitability of AI and confused it with the readiness of every AI product currently being sold. Those are two very different things. The internal combustion engine was inevitable.
That did not mean every man with a wrench in 1905 knew how to build a transmission.
The internet was inevitable.
That did not mean Pets.com was secretly a good business. Smartphones were inevitable.
That did not mean the Palm Foleo deserved to live. Artificial intelligence is inevitable.
That does not mean the voice system at the Taco Bell drive-through should be trusted with the full intellectual burden of "one Crunchwrap, no tomatoes."
According to the video Asmongold was reacting to, companies are already discovering this distinction the expensive way. Businesses rushed AI into scheduling, medical documentation, customer service and other workflows, only to learn that unreliable automation can create more work than it removes. Employees wind up checking the machine, correcting the machine, retraining the machine and apologizing for the machine—all after management froze hiring because the machine was supposedly going to replace them.
That is not an AI problem. That is a management problem wearing an AI conference badge. And Asmongold sees it immediately. These companies did not implement artificial intelligence because they carefully mapped a process, understood the technology, measured the failure modes and designed human oversight around it.
They implemented AI because somebody in the executive meeting said "AI" often enough that everyone else became afraid to admit they did not know what he meant.
Then they fired Debbie from accounts payable. Six months later, the AI is scheduling a forklift operator for outpatient surgery, Debbie is working somewhere else, and the remaining staff are spending all afternoon auditing a machine that was supposed to eliminate their workload. Progress.
We Are Only Going to Die From Our Own Arrogance
While I was watching this, I kept hearing an old Sublime song in my head:
"We're Only Gonna Die for Our Own Arrogance."
It begins with early man walking away and modern man taking control.
That is the whole AI story in one line. Modern man takes control. Modern man builds the machine. Modern man becomes so impressed with the machine that he fires the people who understand the work. Then modern man discovers that owning a machine is not the same thing as understanding one. And that is the real danger here.
It is not that AI is useless. It is that AI is incredibly useful. Useful enough to inspire confidence. Useful enough to work correctly most of the time. Useful enough that people stop checking. That is where it becomes dangerous. A system that fails constantly is easy to reject. A system that succeeds 95% of the time is far more seductive—especially when nobody knows which 5% is fabricated. That is the part too many people still do not grasp.
An AI system can produce something that looks polished, sounds confident, follows the requested format and is completely wrong. And unlike a visibly broken machine, it may not flash a red warning light. It will not lean over and say, "Just so you know, Rich, I made up paragraphs three and six because the statistical vibes felt right." It will simply hand you the document. That means the human reviewing it cannot merely be present.
The human must be competent. You need someone who knows the subject well enough to recognize what does not belong. This is why the fantasy of firing all the experts and leaving one junior employee to "manage the AI" is so profoundly stupid.
The less expertise you retain, the less capable your organization becomes of recognizing machine-generated errors. You have not eliminated the need for expertise. You have eliminated your ability to detect when expertise is missing. That is not transformation.
That is corporate self-harm with a chatbot attached.
The Great AI LARP
There is a lot of AI LARPing happening right now. People are playing the role of advanced AI strategist without having done the deeply unglamorous work of understanding systems, data, operations, incentives, failure, security or actual human behavior. They know the language. They have the slides. They can say "agentic workflow" without blinking. They have a diagram containing six glowing circles, three arrows and the word "orchestration."
Nobody knows what the diagram means, including the person presenting it. But it is blue, so funding is approved. This is how organizations wind up trying to automate processes they never properly understood in the first place.
AI does not repair a broken process simply because you insert it into the process.
Sometimes it allows the broken process to fail faster. Sometimes it scales the error.
Sometimes it hides the error beneath dramatically better prose. And sometimes it works brilliantly. That is the maddening part.
Asmongold's point—and I agree with him—is that the implementation matters more than the slogan. The companies succeeding with AI are not merely spraying it across every department like technological Febreze. They are identifying narrow, valuable pain points. They are selecting the right tools. They are retaining domain experts. They are measuring outcomes. They are building systems around what the technology can actually do rather than what someone promised during a keynote. In other words, they are treating it like engineering. Everyone else is treating it like a costume party.
Why Asmongold Works
Now, part of Asmongold's appeal is that he says things most people only hear from the small, badly supervised creature living in the back of their own minds.
We all have that voice. We just generally do not give it a microphone and 4.65 million subscribers. Asmongold does. He will hear that someone accidentally ordered hundreds of dollars in chicken nuggets and immediately question whether it was really an accident.
Is that fair? Probably not. Is it funny? Unfortunately, yes.
He says the rude first thought. The socially unacceptable thought. The thing your better self catches halfway between your brain and your mouth and wisely sends back for review. His thought escapes. Every time. That is part of the magic. He appears almost physically incapable of allowing a ridiculous statement to pass by without stabbing it with something. But beneath the jokes, there is usually a surprisingly coherent worldview. He understands incentives. He understands that workers often oppose AI because it threatens their livelihoods. He understands that creators sometimes exaggerate public hatred of AI because opposition performs well inside their own communities. He understands that companies chase trends they do not comprehend.
He understands that technological winners emerge from an enormous graveyard of failed products, failed companies and bad implementations. He understands that automation is not new and that countries or companies refusing productivity-enhancing technology may be crushed by competitors willing to use it. And he understands the part many allegedly serious commentators miss: AI is not the fundamental social problem.
The fundamental problem is that human survival remains tied to employment while technology steadily reduces the amount of human labor required to produce goods and services. You can disagree with his answer. You cannot honestly say he has not identified the question.
This is a man making jokes from a room that looks like it may contain several undiscovered biological species, and he is somehow conducting a more substantive labor-economics discussion than half the panel shows on television. Mad respect.
AI Has Reached the Asmongold Layer
Every major technology passes through layers. First comes the research layer. Scientists, engineers and academics talk to one another in language nobody outside the room understands. Then comes the investment layer. Money arrives. Everything becomes revolutionary. Then comes the corporate layer. Consultants build presentations explaining how the revolution will reduce headcount. Then comes the consumer layer. Ordinary people begin using the technology for useful things, stupid things, funny things and occasionally horrifying things. And finally, apparently, comes the Asmongold layer.
This is the moment when the technology has become sufficiently important, sufficiently absurd and sufficiently embedded in the culture that a long-haired man on the internet can pause a documentary, insult nearly every major participant, make several jokes that would get an HR department evacuated, and still land on the correct strategic conclusion:
AI is probably going to change everything. That does not mean you should deploy it like an idiot. Honestly, this may be one of the healthiest stages of technological adoption. The laboratory tells us what is possible. The corporation tells us what is profitable. The investor tells us what is inevitable. Asmongold asks why the AI put bacon in the ice cream. We need that layer. We need people willing to laugh at the emperor while everyone else is still complimenting the tailoring. We need people willing to say that some AI criticism is performative nonsense while also saying that some AI deployments are reckless nonsense. We need enough intellectual flexibility to hold two ideas at once:
AI is enormously powerful. And most people implementing it have no freaking idea what they are doing. Both can be true. In fact, both are clearly true.
Take Your Time
I am not arguing that we should slow AI research to a crawl. The rest of the world is not going to pause while America holds a six-year committee hearing on whether automation makes people uncomfortable.
This technology will advance. It will be deployed. It will eliminate jobs, create jobs, reorganize businesses, reshape education and force us to revisit assumptions about labor, income and human value that have existed for generations. We do not get to opt out. But we do get to decide whether we approach it with humility. We can recognize that a model producing an impressive result does not mean it understands the result.
We can recognize that a successful demonstration is not the same as a reliable production system. We can recognize that human oversight means more than keeping one exhausted employee around to click "approve." We can recognize that expertise becomes more valuable, not less valuable, when machines generate enormous volumes of plausible output. And we can stop pretending that deleting four salaries automatically constitutes innovation.
This is powerful technology. Barely understood technology. Technology moving faster than the organizations trying to govern it and, in many cases, faster than the people trying to sell it. So perhaps Sublime had it right. We are only going to die from our own arrogance. We might as well take our time. Not because AI is going away. Because it is not. Not because it is weak. Because it is becoming extraordinarily powerful. And not because Asmongold is wrong. The long-haired crazy son of a bitch is mostly right.
AI has reached the Asmongold layer. And somehow, against all reasonable expectations, I find that deeply reassuring.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.
Comments