Handmade

There’s something about humans making things. First by hand and then by making ideas in our heads and bringing them to life. In that process we find joy and fulfillment, and most of the time we are driven by curiosity.

For most of history, production worked the same way. You absorbed a tremendous amount of information, and then you added a small thing on top. Something that didn’t exist before, and then you created. Math, science, literature, philosophy, the same pattern in every field where humans excelled.

It wasn’t easy. It took years. Sometimes a whole life. Some people made that contribution. Some didn’t. That’s fine. Everyone contributes the best way they can.

What changed in the past few years is that it’s no longer sure how much of what we make is 100% human, and how much is 100% AI-generated, and how much is in between.

Nothing is wrong with that. Humans used tools throughout history to make better things. The computer in your hand is hundreds, maybe thousands of years of human contribution stacked on top of each other. It started small: realizing we exist, marking things down, counting, a little more arithmetic. Then, over the past 100 or 200 years, the pace picked up, and now we hold machines that do billions of operations per second. That’s production. That’s what all those small contributions add up to.

Moving forward, this human production will be AI-assisted. That’s great, it lets us do more, and faster. But at the same time it takes something away: the ability to produce something entirely by yourself, by hand and by mind.

── ✾ ──

In 2016 I was doing research in Applied Mathematics. I spent hundreds of hours reading references, understanding the topic, finding the main questions, talking with researchers, building relationships, and then making my own contribution and writing my own paper. I was proud of it. It gave me fulfillment. That small piece of work probably had no impact on humanity. But it was a step in the long process of building on each other, one step at a time.

Do the same project today and the experience is completely different. The hours I spent debugging code, Codex would write most of it. Maybe 99%. There are corner cases, but recent models seem to generalize well, and with a verifiable objective they just keep running until they figure it out. The questions I spent weeks answering, a few sessions of asking the model about the papers would answer. Any confusion clears up after a few sessions.

We still get to the solution. But did my brain benefit from the 2016 version? The sleepless nights and the hard work. I think that process contributed to making me more resilient, more mature, better at sitting with a problem I didn’t understand. I don’t think I’d get that from the 2026 version.

But from a pragmatic point of view, the 2026 version solves the problem in hours instead of months, and leaves me time for things I didn’t have time for.

Which would I choose? Hard question.

I like manual work. There’s a reason handmade things cost more than machine-made ones. Not because the quality is better, it’s often worse. It’s because a person spent time on it. Years to master a craft, and then the hours it took to make this one thing. We pay the extra out of respect for that. We know someone made this, with their hands.

So at some point you have to ask: does it make sense to move faster, or does it make sense to enjoy the journey? If a better model lets you produce ten times what you produced yesterday: ten times the code, ten times the research, does that actually benefit you? Maybe in medicine or biotech it does. Maybe it just creates more information than humans can absorb or use. Maybe we don’t need to understand it and other AIs consume it, build on it, and the problem gets solved without a human reading a line. That’s great. But there are things in life we enjoy doing. If you take those away, what’s left? No one knows.

What will probably stay is manual work in the physical world: plumbers, HVAC technicians, jobs that need technical craft and touch real things. Even with robotics coming, I think we’ll still prefer humans there. If my dentist were a robot, I’d question that. If a human dentist uses a precise robot to work on my teeth, I’d be happier, because I’d know a person is in control of the machine. I think this hybrid (human in control) machine doing the hard part is what lets us keep enjoying our work, our existence, our life’s mission, while still using the best tools.

── ✾ ──

So back to the first question: what is this for us? Is it the end of certain jobs, certain skills? The end of math as we know it, of software engineering? No one has an answer. Things are moving faster than we can comprehend the impact. At the surface, people lose jobs. Below that, social structures change, society changes as we know it. What we feel work is for, where fulfillment comes from, what gives a life purpose, those may shift in ways we can’t see yet.

Those questions are complicated, and no one can save you. Don’t expect other people to do things for you. The best thing to do as a human is to understand that this is happening, and then find out what you care about, what you like to do, and spend time on it.

It’s fine to write code by hand. It’s fine to let AI write code for you. It’s fine to do math research by hand, and fine to do it with AI. What matters is a clear distinction between what’s human and what’s machine-generated.

A “human made” label is like a “handmade” label on a t-shirt. It doesn’t make the t-shirt more comfortable. It’s still a t-shirt. But you know there’s a person behind it. That label is based on trust, and it’s a good thing to have for future work. For what already exists, a date is probably a good label (pre-2022 or not). Nothing will change people’s minds about a paper written by hand in 2020 versus a nicely polished project written by AI in 2026. The AI code is probably more reliable and works better. But the human code is beautiful to see, because it proves humans can actually do complicated things. And all of this AI today is itself a human production.

What’s going to happen to us? No one knows. What’s going to happen to the human brain? No one knows. There are early signs that people won’t just sit back and accept changes from AI, that we’ll need to find fulfillment somewhere else.

Keep doing things by hand. Keep using AI. Don’t forget how to do things by hand when needed. It’s fun to actually make something, and build, and contribute.

Note: This essay was transcribed by AI from recorded audio notes, but the ideas and examples are given verbatim from the author.