The Kitchen, the Code, and the Shift Nobody Sees Coming
An essay on what AI actually is, where it's going, why most people are preparing for the wrong future, and what happens when you stop using AI as a tool and start rebuilding your entire way of thinking around it.
Picture two kitchens.
Same ingredients. Same recipe book. Same stove, same knives, same clock on the wall. If you walked in and looked around, you'd say they were identical. And you'd be right, on paper.
In Kitchen A, the cook follows the recipe step by step. A little chopping here, a pinch of salt there. Sometimes he forgets the garlic. When something burns, he's not entirely sure why. The result? Edible. Sometimes good. Sometimes just fine.
In Kitchen B, same ingredients, same recipe. But something is different. This cook reads the recipe and immediately breaks it apart, not as instructions, but as a system. She knows that searing requires a hot pan before the oil goes in. She tastes as she goes. She adjusts. She improvises. She doesn't just follow the recipe she understands it.
Both kitchens produce food.
But only one produces a chef.
This, in essence, is the difference between how most people are using AI today, and how a small minority will use it tomorrow. Most are in Kitchen A. They're treating AI as a tool, a faster blender, an auto-pilot, a magical assistant that spits out summaries or writes their emails or generates code they don't quite understand. And that works. For now.
But the ones who will actually thrive in the next decade aren't the ones asking, "How do I use AI to do my job faster?"
They're the ones asking, "How does AI change what my job even is?"
That shift from using AI as a tool to rebuilding your thinking around AI is the real transition. And almost no one is making it.
Part I: What AI Actually Is (And Why Most Explanations Miss the Point)
Let's start with the boring truth: most people who talk about AI don't actually know what it is. Not because they're dumb because it's genuinely hard to explain, and the metaphors we use are usually wrong.
The most common metaphor is: AI is a smart assistant. It helps you. It answers questions. It does tasks. It's like a really good intern who never sleeps.
This metaphor is useful, but dangerously limited. Because it implies a clear division: you are the thinker, and AI is the doer. You decide, it executes. You ask, it answers. The mental hierarchy stays intact.
But that's not what's happening.
What's actually happening especially with large language models is something stranger. AI isn't a tool that executes your commands. It's more like a mirror that reflects cognition back at you, but in ways that often reveal things you didn't know you were thinking.
Let me explain.
When you write a prompt, you're not just issuing an instruction. You're encoding your assumptions, your framing, your blind spots, your priorities, your style of thought. The model doesn't just answer your question it responds to the shape of your mind as expressed through your words.
That's why two people can ask "the same question" and get wildly different results. It's not that one is better at prompting. It's that prompts are thoughts made visible, and different people think differently.
This is what most people miss. They think the skill is "writing good prompts." But the real skill is thinking clearly enough that your prompts become precise.
AI doesn't make you smarter. It makes your thinking legible which is different, and in some ways harder.
Part II: Why Most People Are Preparing for the Wrong Future
There's a common narrative about AI and the future of work. It goes like this:
"AI will automate routine tasks. Humans will move up the value chain. We'll focus on creativity, strategy, and emotional intelligence the things machines can't do."
This sounds reasonable. It's also almost entirely wrong.
Not because AI won't automate routine tasks it will. But because the assumption that humans will "move up" misunderstands how displacement actually works.
Here's what usually happens:
1. A new technology automates part of a job.
2. The remaining parts become more competitive.
3. People assume they'll be fine because they do the "higher-level" stuff.
4. Then the technology gets better and starts doing that too.
5. Repeat.
The flaw in the "move up the value chain" narrative is that it assumes the chain is stable. It assumes that the categories we use today "strategy," "creativity," "management" will still mean the same thing in five years. They won't.
AI doesn't just automate tasks. It redefines what tasks are valuable.
Think about writing. A few years ago, the value chain looked like this:
Now look at what's happening:
If everyone can generate "thought leadership" with AI, then "thought leadership" isn't the differentiator anymore. The differentiator becomes taste the ability to know what's good, what's relevant, what's true. And taste is something AI can mimic but not originate.
So the future isn't about "moving up." It's about moving sideways into capabilities that are orthogonal to the automation curve. Not "higher" value, but different value.
Part III: The Real Skill Is Not Prompting It's Thinking in Loops
Let's go back to the kitchen.
In Kitchen B, the chef doesn't just follow the recipe. She thinks in loops:
This is called iterative reasoning. And it's the single most important skill for working with AI effectively.
Most people use AI in a straight line:
1. Ask a question.
2. Get an answer.
3. Use the answer.
That's fine for simple tasks. But for anything complex writing, research, strategy, code this approach fails. Because the first answer is almost never the best answer. It's a starting point.
The people who get the most out of AI are the ones who treat every output as a draft. They don't accept; they interrogate. They ask follow-up questions. They push back. They say, "That's not quite right try again with this constraint." They use AI not as an oracle, but as a sparring partner.
This changes the nature of the interaction. It's no longer "ask and receive." It's "propose, critique, refine, repeat."
And here's the key insight: this loop isn't just about getting better answers. It's about training yourself to think more clearly.
When you have to articulate why an answer isn't quite right, you're forced to clarify your own standards. When you have to describe what "better" looks like, you're forced to define your goals. When you have to break a vague request into specific instructions, you're forced to confront your own ambiguity.
AI doesn't do the thinking for you. But it forces you to externalize your thinking, which is often the hardest part.
Part IV: The Coming Divide
Here's what I think is going to happen over the next five to ten years.
There will be a split. Not between "people who use AI" and "people who don't." Almost everyone will use AI in some form it'll be as ubiquitous as Google is today.
The split will be between:
1. People who use AI as a faster way to do the same things. They'll automate tasks, save time, maybe get a little more done. But their fundamental way of thinking won't change. They'll be in Kitchen A competent, efficient, replaceable.
2. People who use AI as a way to think differently. They'll use it to challenge assumptions, explore alternatives, stress-test ideas, externalize cognition. They'll iterate, interrogate, refine. They'll be in Kitchen B creative, adaptive, irreplaceable.
The first group will compete on speed. The second group will compete on depth.
And here's the uncomfortable truth: most people will end up in the first group. Not because they lack the ability, but because they lack the awareness. They'll learn "how to use AI" the way they learned how to use Excel as a skill to acquire, a box to check, a line on a resume.
They won't realize that the real opportunity isn't to use AI better. It's to become a different kind of thinker one who treats AI as an extension of cognition, not a substitute for it.
Part V: What It Looks Like to Actually Do This
Let me get concrete.
I've spent the last two years rebuilding the way I work around AI. Not "using AI more." Rebuilding. Here's what that looks like in practice:
1. I write with AI, not using AI.
When I write an essay like this one, I don't start with a blank page. I start with a conversation. I tell the AI what I'm trying to say roughly, imperfectly, often in fragments. Then I argue with it. I say, "That's too abstract." Or, "Give me a better metaphor." Or, "What's the strongest counterargument to this?"
The AI doesn't write the essay. But it shapes the essay by forcing me to articulate what I actually mean.
2. I use AI to find the holes in my thinking.
Before I commit to a decision business, creative, personal I run it through AI with a specific prompt: "Argue against this. Tell me what I'm missing. Be harsh."
This isn't about getting "the right answer." It's about stress-testing. I want to see the weaknesses before they become problems. AI is brutally good at this if you ask it to be.
3. I treat AI as memory augmentation.
I have a system where I dump notes, ideas, fragments, half-baked thoughts into a running conversation. Then, weeks later, I can ask: "What have I been thinking about lately?" or "Is there a connection between these three ideas?" or "What was that insight I had about X?"
AI becomes external memory a way to retrieve and recombine thoughts that would otherwise be lost.
4. I never accept the first answer.
This is the simplest habit, but also the hardest to build. Whenever AI gives me an answer, I assume it's a draft. I push back, refine, iterate. Usually, the third or fourth version is dramatically better than the first. But most people never get there because they accept the first response and move on.
Part VI: The Philosophical Shift
There's a deeper layer here, one that most discussions of AI avoid because it gets uncomfortable.
AI forces us to confront a question we've been dodging for centuries: What is thinking, really?
We like to believe that thinking is something uniquely human. That consciousness, creativity, and insight are special properties that can't be replicated. And maybe that's true in some ultimate sense.
But what AI reveals is that a lot of what we call "thinking" is actually pattern-matching, recombination, and prediction. We take inputs, we process them according to learned rules, and we produce outputs. We do this so automatically that we call it "intuition" or "judgment" or "creativity." But underneath, there are patterns.
This isn't a reason to feel diminished. It's a reason to feel curious.
If AI can do some of what we thought only humans could do, then maybe we've been wrong about what makes us human. Maybe the interesting part isn't the pattern-matching it's what we do with the patterns. The choices we make. The values we hold. The questions we ask.
AI can generate text. It can't decide what's worth saying.
AI can optimize. It can't decide what's worth optimizing for.
AI can answer questions. It can't decide which questions matter.
The more AI handles the mechanical parts of cognition, the more humans are pushed toward the philosophical parts meaning, purpose, judgment, ethics.
That's the real shift. Not "humans vs. machines." But humans becoming more distinctly human by offloading the less-human parts of thought.
Part VII: What to Do Now
If you've read this far, you might be wondering: "Okay, but what should I actually do?"
Here's my advice, as concretely as I can put it:
1. Stop thinking of AI as a tool. Start thinking of it as an environment.
A tool is something you pick up, use, and put down. An environment is something you inhabit. The difference matters because it changes your relationship. You don't "use" an environment you adapt to it, you navigate it, you shape it and are shaped by it.
AI is becoming an environment. The question isn't "How do I use this tool?" It's "How do I live and work in this new environment?"
2. Develop your taste.
In a world where anyone can generate content, the bottleneck shifts to curation. Who can tell what's good from what's mediocre? Who can see what's missing? Who knows when to stop iterating and when to push further?
Taste is judgment. It's pattern recognition at a higher level not "what's next in the sequence" but "what's worth doing in the first place."
3. Practice articulating your thoughts.
The better you can express what you're thinking, the more effectively you can work with AI. This isn't about "prompting." It's about clarity. Can you say, in words, what you want? Can you explain why something isn't working? Can you define your criteria?
Most people are surprisingly bad at this not because they can't think, but because they've never had to externalize their thinking with precision.
4. Learn to iterate.
The people who get the most out of AI are the ones who treat every interaction as a loop, not a transaction. They don't ask once and accept. They ask, refine, push back, ask again. They treat AI like a collaborator who needs direction, not an oracle who delivers truth.
5. Stay philosophical.
The technical side of AI is changing fast. But the human side what it means to think, to work, to create, to decide changes slowly. If you ground yourself in those deeper questions, you'll be more resilient to whatever shifts come next.
Epilogue: The Kitchen, Revisited
Back to the two kitchens.
Both cooks have access to the same recipe, the same ingredients, the same tools. But one is following instructions, and the other is understanding principles.
AI is about to flood every kitchen with new equipment faster, smarter, more capable than anything we've seen. Some people will use that equipment to follow instructions faster. Others will use it to deepen their understanding.
The equipment doesn't decide which one you become.
You do.
Thole, April 2026