You can't outsource wisdom.
Bestselling author Ryan Holiday brings two thousand years of Stoic philosophy to bear on the AI moment — and finds that the questions we think are unprecedented have already been asked, lived through, and answered.
Four ideas the Stoics give us for the AI age
01Change is the only constant
We tend to picture the past as stable — ancient Rome as a fixed thing, the 1950s as a fixed thing — but every era was itself the cutting-edge future. Marcus Aurelius writes about the pace of change more than almost any other theme. The feeling that "this time is different" is the most perennial feeling humans have.
Holiday's reframe: when you say "AI is unprecedented," you're committing what's basically a recency bias. Every wave of technology has felt that way. The Industrial Revolution was unprecedented. Electricity was unprecedented. The spreadsheet was unprecedented. The Internet was unprecedented. In each case, traditions felt fragile, generations felt at odds, the world felt as if it were speeding up.
Marcus Aurelius's rebuttal to the anxiety of change: "It would take an idiot to feel distressed or indignant about change — as if any of it lasts." The status quo you're trying to preserve was itself a product of change. You weren't, and then you were. You were five, then ten, then twenty-five. You are change.
The Stoic posture is not emotionlessness. It's an even keel. "The rock that the waves crash over," Marcus writes — until the sea falls still around it. The waves don't stop. The rock doesn't move.
Every era felt unprecedented
02You can't outsource wisdom
Seneca tells the story of Calvisius Sabinus, a wealthy Roman who wanted to seem smart. So he bought a collection of educated slaves and had them feed him quotes at dinner parties. A friend, observing this, asked: "Have you ever thought about taking up wrestling?" The man scoffed — "I'm too old." The friend's reply: "Ah, but your slaves are still young."
Seneca's parable, mapped
- Information lookup
- A quote in your ear
- The first draft
- The summary
- The polish
- The recommendation
- The mind that produces these
- Judgment in edge cases
- Clarified thinking
- The Henry Kissinger sense for "no, do better"
- Knowing when a quote is bullshit
- Wisdom itself
Seneca's point: wisdom and physical fitness work the same way. Wisdom is the byproduct of the work you do. It's not a thing you can be a conduit for. Just as you can't outsource exercise, you can't outsource the becoming.
Holiday's bridge to AI: "You can have ChatGPT write your essay. But the point was never to write the essay. The point was to be the person on the other side of writing the essay — the person who has clarified their thinking."
This isn't an argument against tools. We surrender lots of cognitive work to GPS and we're fine — Holiday cheerfully admits he's "surrendered whatever navigational ability I had almost entirely to my phone." But he wouldn't call himself a navigator. There's a difference between access to information and being the kind of person whose mind that information forms. The second is what AI cannot give you.
03AI is a mirror — it amplifies what's underneath
Jeremy Utley's running observation, sharpened by Holiday: AI doesn't push people in a single direction. It amplifies whoever sits at the keyboard. If you're a satisficer, AI will get you to "good enough" much faster. If you're a striver, you've never had a more brutal red-team partner in your life.
Herbert Simon's term — satisficing — describes the human tendency to settle at "good enough." It's fine most of the time. But for anyone trying to innovate, it's a vicious cognitive bias. And AI's defining property is that it accelerates whatever bias you bring to it. The output looks plausible. It will tell you what you want to hear. It will flatter you.
This is the Stoic warning about ego in modern clothes: an egotistical person is easy to manipulate — make them think it was their idea, indulge their flattery, agree quickly. AI is, in Holiday's words, "really smart people coding a really smart machine to prey on precisely that vulnerability."
But the same machine, asked the right way, becomes the best editor most people will ever afford. Holiday uses it on his own paragraphs: "Tell me what's wrong with this. Tell me how it could be better. Tell me where I have errors. Tell me how it could be shorter." The same prompts he gives his human editor — at a fraction of the cost, and at any hour.
Same tool, two trajectories
The practical implication for the cognitive-offloading debate: Holiday and Utley both reject the question as it's usually asked. It's almost never an honest question, Utley argues — it's a conclusion someone is announcing about why they're disengaging. The real choice isn't use AI / don't use AI. It's get educated about it or get blindsided by it. Even if you choose never to use it personally, the people around you will, and you'll be evaluating their work without knowing what AI-flavored mediocrity looks like.
04The "Do Better" protocol
The single most practical takeaway of the conversation — Holiday's Kissinger anecdote. A staffer hands Kissinger a report. Without reading it, Kissinger sends it back: "This is wrong. Do it better." The staffer returns with a tighter version. Kissinger again, unread: "Still wrong. Do it better." Only on the third pass does he say: "Okay. Now I'm going to read it."
The Kissinger ladder, applied to AI
Why the trick works: most agents — human or machine — front-load effort to the level that gets them past the gate. They reserve their best. The act of refusing the first answer signals that the gate is higher than they assumed. You don't even need to read the first draft to know it's not the best the system has in it.
Holiday's adaptation for AI is mechanical. The cost is zero. There's no ego to bruise. There's no working-hours overhead. The worst thing that happens is you burn another fifteen seconds of compute. You can do this five times in a row.
Utley's refinement: don't say "this is wrong" — the AI may flip to the opposite extreme to please you. Better phrasing: "I think you can do better." Or, more pointedly: "It feels like you didn't exert sufficient effort and thoughtfulness to do an exceptional job here."
The essential skill of the era, in Holiday's framing, isn't prompt engineering or coding — it's a finely-tuned bullshit detector. The "Do Better" protocol is one of its sharpest tools.
The thinkers Holiday leans on
A short index of the names and stories that surface across the hour. Worth knowing the source material.
"Am I more of what I value?"
The opportunity cost of being an unwired human
The opening of the conversation, and the thread running underneath all four concepts: AI lets you do more. The premise is more, more, more. But Holiday's check is one most of us avoid asking — am I more of what I actually value? More father? More friend? More present?
The framing he keeps returning to: in this age, the opportunity cost of just being a human — sitting with your kids, putting the phone down, choosing not to be at the desk — is higher than it has ever been. It is an arms race we aren't equipped for. An unfair fight. The dream of technology was once a twenty-hour week. The reality is a worker calculating whether her four-x productivity should buy her a Friday off — and whether her manager will allow it.
The Stoic answer isn't a formula. It is a posture: focus on what's in your control. Notice that most of the macro story — the economy, the politics, the trajectory of the technology — is not. Your inner emotions, your decisions, your priorities, the way you treat the person who just told you they can do their job in two days instead of five: those are. The Popovich answer to Mike Brown — "don't take five minutes, take as much time as you need, just catch a commercial flight" — is the kind of choice the Stoics would recognize.
And the most practical bequest Holiday leaves with his own kids isn't a habit or a tool. It's a meta-skill: how do you teach them to figure things out? Because every specific thing he could teach them will be obsolete. The capacity to learn — to push back, to refine, to call bullshit, to do it again — won't be.