I turn to see our pouting three-year-old enter the room, expectantly awaiting retribution.

“She’s copying you?” I ask.

She nods.

I consider this.

“You know,” I tell her. “Some people say that copying is the biggest way to compliment someone. She must really like what you’re doing.”

She pauses just a moment, then turns on her heels and immediately back down the stairs:

“Okay, Sissy! You can copy me!”

“Okay, Sissy! You can copy me!” I hear her big sister parrot back from the floor below.

Game. On.

It’s a tiny moment, but it captures something fundamental about how humans learn. Way before school, way before careers, and certainly way before AI.

When I think back on my own career, all of my most pivotal moments in my trajectory have been inspired by seeing someone else do it first. In a way, artificial intelligence has only exponentially amplified what humans (even 3- and 5-year-olds) have known all along: We copy to integrate. We copy to learn. We copy to grow.

Notably, the most effective way I’ve seen people pick up AI isn’t new at all. First, they see what someone else is building. Then, they find a reason to build something of their own. Artists, writers, and musicians who emulate each other’s styles in their own work have known this for years: Copying isn’t a shortcut to creativity; it’s the on-ramp.

Interestingly, this same logic applies to the models themselves. Large language models were trained by absorbing our collective knowledge and formalizing it in code. When we ask AI to act, we usually ask it to step into the role of someone else (ie: a chef, a teacher, or a coach).

Two years ago, it never would have occurred to most of us to extend this mirroring mode to the extreme ad ask the AI to literally assume the identities of other people. But recently, Markup AI’s CEO Matt Blumberg did just that. He and his team created an “AI Fantasy Board of Directors” made up of public figure personas like Warren Buffet, Oprah Winfrey, and Steve Jobs.

In addition to borrowing ideas from people we admire, the way we are learning AI is also pushing learning out of the classroom in the most significant way since the modern school system itself. We aren’t going back to school to learn AI. We’re learning it, anyway.

Instead, we’re staying up late. We’re hacking with friends. We’re following a new class of builder-turned-teachers on YouTube, Twitter, Substack, and LinkedIn.

Engineering is becoming social again. And anyone who shares their builds in public is participating in a new kind of coding commons.

Watching the evolution of the newly mainstreamed Clade Code has reminded me that the ceiling of these seemingly transcendent tools is defined less by what they can do than by the limits of our own imagination.

Six months ago, I felt like a superhuman non-engineering hero every time I prompted something in the command line interface. But today it seems like everyone in my personal and professional network is moonlighting as an app developer.

Some days I’m even tempted to believe I’m keeping up with my most talented engineering friends (yes, even the ones with decades of experience in app development or CS degrees from prestigious universities).

But when I see those friends post photos of their work setups, showing not just one or two coding instances, but 17 simultaneous monitors, I’m reminded that acceleration has limits. Learning faster doesn’t mean becoming someone else.

The user, Horsefacts posted on Farcaster about their multi-monitor “Claudecave setup that run 17 simultaneous instances of Claude Code

AI isn’t a miracle panacea that creates new personas from nothing. It’s a mirror and an accelerant, intensifying the edges of who we already are. For a 10x software engineer, it can enable 100x output. But even with access to the same tools, I won’t become a 100x engineer overnight.

For someone like me, AI has intensified the part of me that experiments in public, learns by building, and applies the same generalist, hyper-curious nature across all of my personal and work projects.

If we’re all thrust into this new classroom at the same time, even head start gains from early adopters will eventually normalize over time as embedded experts learn the same tools. Which means the real question is what you do with your edge. The so-personal-it-can-only-come-from-you bit that AI helps you unlock.

In my case, I feel that AI has given me a rare do-over and a chance to return to project themes I’ve carried for years and finally see them through, now that I have the tools. What will you do with your edge?

I read a fantastic essay this week where Abi Awomosu posits (among many other useful things) that Big Tech companies didn’t invent AI; they simply discovered the rules that govern its existence.

In the post, she likens the behaviors of artificial intelligence to those of other radical “attention technology” breakthroughs. She outlines everything from the origin of speech and language to the industry revolution, the arrival of the internet and now, of course, the AI age. Here’s a handy chart (AI-generated, of course) that summarizes this thesis:

What I like about this framing is that it’s positioning AI not as a tool, but as a novel new human intelligence and communication framework. It just so happens to be a framework we’re all getting introduced to around the very same time.

And in that way, we’re all in the same moment of learning the same way anyone learns something new: Through mirroring, experimentation, and play.

When I think back to this weekend morning, playing copycat games with my kids, what strikes me isn’t the copying. It’s the permission.

My daughter didn’t stop being herself when her sister copied her. She didn’t lose ground. She realized that copying was how the game began…not how it ended.

That’s what AI is offering us, too. An invitation to start. What we build after that is still up to us.

I turn to see our pouting three-year-old enter the room, expectantly awaiting retribution.

“She’s copying you?” I ask.

She nods.

I consider this.

“You know,” I tell her. “Some people say that copying is the biggest way to compliment someone. She must really like what you’re doing.”

She pauses just a moment, then turns on her heels and immediately back down the stairs:

“Okay, Sissy! You can copy me!”

“Okay, Sissy! You can copy me!” I hear her big sister parrot back from the floor below.

Game. On.

It’s a tiny moment, but it captures something fundamental about how humans learn. Way before school, way before careers, and certainly way before AI.

When I think back on my own career, all of my most pivotal moments in my trajectory have been inspired by seeing someone else do it first. In a way, artificial intelligence has only exponentially amplified what humans (even 3- and 5-year-olds) have known all along: We copy to integrate. We copy to learn. We copy to grow.

Notably, the most effective way I’ve seen people pick up AI isn’t new at all. First, they see what someone else is building. Then, they find a reason to build something of their own. Artists, writers, and musicians who emulate each other’s styles in their own work have known this for years: Copying isn’t a shortcut to creativity; it’s the on-ramp.

Interestingly, this same logic applies to the models themselves. Large language models were trained by absorbing our collective knowledge and formalizing it in code. When we ask AI to act, we usually ask it to step into the role of someone else (ie: a chef, a teacher, or a coach).

Two years ago, it never would have occurred to most of us to extend this mirroring mode to the extreme ad ask the AI to literally assume the identities of other people. But recently, Markup AI’s CEO Matt Blumberg did just that. He and his team created an “AI Fantasy Board of Directors” made up of public figure personas like Warren Buffet, Oprah Winfrey, and Steve Jobs.

In addition to borrowing ideas from people we admire, the way we are learning AI is also pushing learning out of the classroom in the most significant way since the modern school system itself. We aren’t going back to school to learn AI. We’re learning it, anyway.

Instead, we’re staying up late. We’re hacking with friends. We’re following a new class of builder-turned-teachers on YouTube, Twitter, Substack, and LinkedIn.

Engineering is becoming social again. And anyone who shares their builds in public is participating in a new kind of coding commons.

Watching the evolution of the newly mainstreamed Clade Code has reminded me that the ceiling of these seemingly transcendent tools is defined less by what they can do than by the limits of our own imagination.

Six months ago, I felt like a superhuman non-engineering hero every time I prompted something in the command line interface. But today it seems like everyone in my personal and professional network is moonlighting as an app developer.

Some days I’m even tempted to believe I’m keeping up with my most talented engineering friends (yes, even the ones with decades of experience in app development or CS degrees from prestigious universities).

But when I see those friends post photos of their work setups, showing not just one or two coding instances, but 17 simultaneous monitors, I’m reminded that acceleration has limits. Learning faster doesn’t mean becoming someone else.

The user, Horsefacts posted on Farcaster about their multi-monitor “Claudecave setup that run 17 simultaneous instances of Claude Code

AI isn’t a miracle panacea that creates new personas from nothing. It’s a mirror and an accelerant, intensifying the edges of who we already are. For a 10x software engineer, it can enable 100x output. But even with access to the same tools, I won’t become a 100x engineer overnight.

For someone like me, AI has intensified the part of me that experiments in public, learns by building, and applies the same generalist, hyper-curious nature across all of my personal and work projects.

If we’re all thrust into this new classroom at the same time, even head start gains from early adopters will eventually normalize over time as embedded experts learn the same tools. Which means the real question is what you do with your edge. The so-personal-it-can-only-come-from-you bit that AI helps you unlock.

In my case, I feel that AI has given me a rare do-over and a chance to return to project themes I’ve carried for years and finally see them through, now that I have the tools. What will you do with your edge?

I read a fantastic essay this week where Abi Awomosu posits (among many other useful things) that Big Tech companies didn’t invent AI; they simply discovered the rules that govern its existence.

In the post, she likens the behaviors of artificial intelligence to those of other radical “attention technology” breakthroughs. She outlines everything from the origin of speech and language to the industry revolution, the arrival of the internet and now, of course, the AI age. Here’s a handy chart (AI-generated, of course) that summarizes this thesis:

What I like about this framing is that it’s positioning AI not as a tool, but as a novel new human intelligence and communication framework. It just so happens to be a framework we’re all getting introduced to around the very same time.

And in that way, we’re all in the same moment of learning the same way anyone learns something new: Through mirroring, experimentation, and play.

When I think back to this weekend morning, playing copycat games with my kids, what strikes me isn’t the copying. It’s the permission.

My daughter didn’t stop being herself when her sister copied her. She didn’t lose ground. She realized that copying was how the game began…not how it ended.

That’s what AI is offering us, too. An invitation to start. What we build after that is still up to us.