Dan Koe recently published a long essay laying out his vision for what an AI society could look like over the next 5 to 10 years. He says people may gradually split into three groups: “purists” who reject technology, “automated humans” who outsource judgment to AI, and a third type that can keep command between technology and human nature, which he calls the “Digital Renaissance Man.”
In Koe’s framing, the world is entering a “second renaissance.” Resumes will be replaced by work, employers will be replaced by audiences, and taste will increasingly take the place of degrees and certificates.
Generative AI and the rise of the one-person company
Koe argues that the highly specialized career structure shaped after industrialization trained people to own only a small part of the workflow: designers handled design, engineers wrote code, marketers brought traffic, and sales teams closed deals.
Generative AI is now pushing that model in the opposite direction. Coding, design, research, content, marketing and even website development can increasingly be done with AI assistance. Some work that once needed teams of 10 to 50 people may now be handled by one person who knows how to direct AI.
That is why Koe puts forward the idea of a “One-Human Business.” In his view, one person can use social media for free distribution, use AI to write code and build products and websites, then rely on personal expertise to create something differentiated.
He compares the shift with earlier waves of technological replacement. Scribes moved toward editing after printing presses displaced copying by hand. Hand weavers shifted toward operating and managing machines after mechanization. AI, he argues, is automating the most mechanical and repetitive parts of skill, while human value moves toward direction, taste, judgment and decision-making.
The riskiest career moat may be tool-specific skill
Koe breaks what he sees as an “irreplaceable skill stack” into three layers.
The first is understanding human nature. That includes marketing, sales, writing, communication and psychology. No matter how technology changes, he says, people still need to understand attention, demand and what the market will treat as valuable.
The second layer is a person’s own mix of expertise and interests. That may include philosophy, fitness, design, programming, nutrition, finance or any other field. Koe argues that differentiation usually comes not from a single specialty, but from the overlap between domains. Someone with multiple forms of experience and perspective is harder to copy than someone confined to one professional frame.
The third layer is tool and technical execution, such as Photoshop, email marketing tools, website production, social media management or hands-on coding. That, he says, is also the layer AI can erode most easily.
His warning is direct: if a person’s career moat still rests on “I can operate this piece of software better than others,” the risk may keep rising.
Why Koe prefers an apprenticeship model in the AI era
Koe also criticizes the traditional education model of learning first and practicing later. He says the more efficient sequence runs the other way around: try to do the work first, run into problems, then look for the knowledge needed to solve them.
He says that sequence is closer to the old path from apprentice to journeyman to master. Today, people do not even need to physically enter a master’s workshop. Books, YouTube, podcasts, Substack and AI can together form what he describes as an almost unlimited self-education system.
But the key, in his view, is not how much knowledge someone collects. What matters is having a real goal that keeps testing that knowledge against reality. For that reason, even if AI can produce false information, Koe argues that the skill people need is not rejecting AI, but building the capacity to question, verify and experiment.
AI slop, first drafts and iteration
Koe is equally blunt about the flood of what he calls “AI Slop.” He says one of the biggest mistakes people make with AI is treating the first output as the finished product. Whether the task is a website, software, an article or a product, the first version is usually not very good. That was true before AI as well. The real work happens in the dozens or even hundreds of iterations that follow.
He gives three principles for using AI:
- Know what you want, and know what good looks like. When anyone can vibe code an app, AI will replace the people who do not know what quality is.
- Keep standards high. Koe says AI is not a “genius machine” that produces a perfect answer in one shot. It behaves more like a tool that gets closer to the target through the user’s requests, judgment and iteration.
- Try to replace yourself with AI. Turn the work you do each day into prompts, workflows, skills, agents or software processes. Only by trying to automate the work will you learn what AI handles well and what still needs to stay under human control.
Koe uses his own research process as an example. He says AI is well suited to scanning large volumes of YouTube videos, articles and social media posts to find trends and patterns. But he still chooses to write the article itself by hand, because the writing process and the precise control over reader value are parts of the work he wants to keep.
The future worker, in Koe’s view, is a walking contradiction
Koe’s description of future talent is not centered only on “understanding AI.” What stands out more is the ability to hold together traits that look contradictory: using AI deeply while also being able to live away from the screen; producing intensely while also resting fully; understanding business while studying philosophy; creating while also building systems; being both a specialist and a generalist.
He calls that kind of person a “walking contradiction.”
Koe also points to the lifestyle of Charles Darwin to argue that high output does not necessarily mean long working hours. According to Koe, Darwin reportedly spent only about 3 hours a day in truly concentrated creative work, with the rest of the day spread across walking, reading, rest and family life.
Koe says AI may make that kind of work pattern possible again: a person could spend only a few hours each day on creative work that truly requires deep human focus, yet still produce output that once required 10 to 20 people, with AI and automation systems doing the rest.