Crash the UFO in Your Backyard
- Rich Washburn
- 1 day ago
- 8 min read
About four years ago, when I first started speaking publicly about AI, I used to frame it with a thought experiment.
Imagine you get home from work, turn on the news, and every channel is carrying the same breaking story: Aliens are coming.
Not maybe. Not some blurry light over Nevada. They've contacted us. They're real. And they're going to land on Earth in two years.
What do you do? Because whatever your answer is, I promise you one thing: You probably don't keep behaving exactly the way you did yesterday.
You start paying attention. You start reading. Governments start preparing. Scientists lose their minds. Markets move. Religions have some meetings.
And somewhere, somebody immediately starts asking the most interesting question: What do they know that we don't?
That was how AI felt to me then. Something enormous was coming. Something intelligent.
Something that was going to completely change what humans could do.
Well.
Apparently we've moved past waiting for the aliens to land. Now we're crashing the bastards in our backyards. And I think we should start taking them apart.
THE PART PEOPLE ARE MISSING
We talk constantly about what AI lets us produce.
Code.
Websites.
Research.
Applications.
Hardware.
Presentations.
Designs.
Agents.
Automation.
You can now walk into fields where you have very little formal expertise and make things that would've been completely inaccessible to you a few years ago.
I do this constantly... build an ESP32 device... write a desktop application... create a data pipeline...build an agent. I can wire sensors together and write the firmware that makes them work.
There are plenty of people who actually know these disciplines who could absolutely destroy me in a technical conversation about them.
That's not false humility. That's the point.
I can still make the thing. And increasingly, I can make the thing work really damn well. That alone is extraordinary. But I think there's a second unlock hiding behind the first one. And it may matter more.
Once the thing works, take it apart.
I'VE ALWAYS BEEN THIS WAY
I didn't immediately recognize this as a method because, apparently, I've been doing it my whole life. I'm a tinkerer.
When I was a kid, I took everything apart. Everything.
At some point when I was around seven, I was apparently no longer supposed to have access to a screwdriver. Which, in retrospect, seems like a perfectly reasonable policy.
Give me an object and eventually I wanted to know what was inside it. Not because I necessarily intended to fix it. Sometimes the original object did not survive the educational process.
But I needed to know.
Why does that move? What's behind this panel? What happens if I remove that? Where does this wire go? What's this thing actually doing? That wiring never went away.
So when AI helps me build something now, I naturally do the same thing.
I get it working... play with it... break it... change things... ask why something was designed that way... trace pieces backward... start pulling conceptual panels off.
And until recently, I didn't really think of that as anything unusual.
Then I spent more time watching other people use AI. I worked with Andy. I've worked with clients. I've watched technically sophisticated people suddenly realize how much farther they can reach with these tools.
And today I spent some time doing STEM work with a couple of kids.
Somewhere in all of that, something clicked. Maybe the reverse-engineering part isn't incidental. Maybe it's the whole damn thing.
BUILD FIRST
There's a traditional idea of learning that makes perfect sense.
You learn the fundamentals. Then the tools. Then the methods. Then you practice. Then, eventually, you build something sophisticated.
There's nothing wrong with that model. It works. But AI gives us access to another one.
Build first. Make the thing. Get the dopamine hit. See the result.
Create something you didn't believe you could create.
Let the agent write the code. Let it help design the circuit. Let it stand up the environment.
Let it get you across the part where you normally would've spent three weekends reading documentation and swearing at Stack Overflow.
Get all the way to the working object.
Then stop.
That's the moment that matters. Because now there's a UFO sitting in your backyard. And unlike the hypothetical one from my old talks, this one's yours.
Open the hatch.
REVERSE-ENGINEER THE THING
Suppose AI just helped you build a small embedded device.
Great.
Now ask: Why this microcontroller? Why that resistor? Why does this pin need that state? What protocol is carrying the data? What happens during initialization? Where's the state stored? Why does this loop behave that way? What happens if I remove this line? What happens if I change that value? Why did you choose this architecture instead of another one?
And here's the wonderful part: The same machine that helped build the damn thing is sitting right there waiting to explain it.
You don't even need a separate teacher.
Tell it: Explain this piece to me. Now explain it more simply. Now explain it at the electrical level. Now explain what the software is doing. Now quiz me. Now make me predict what'll happen if I change this. Now don't tell me the answer until I commit to one. Now show me where I'm wrong.
That's a very different relationship with AI than simply asking it to make things.
The artifact becomes the curriculum.
THIS IS THE TIME DIVIDEND
People talk constantly about AI saving time.
Good. It does.
But saving time is only valuable if we decide what the saved time is for.
If something used to take twenty hours and AI lets you do it in two, the obvious business instinct is:
Fantastic. Now do ten times as much. And yeah. Sometimes. I like getting shit done too. But I think there's another option we should talk about much more seriously.
Take part of that eighteen-hour dividend and buy understanding with it.
You already got the result. You already skipped the tedious part. You already climbed the hill using a helicopter.
Fine.
Now walk around at the top and figure out where the hell you are. Because if you simply take the result and move on, AI gave you an artifact.
If you stop and dissect it, AI may have given you a new mental model.
Those aren't remotely the same thing.
THE FLYWHEEL
This is where it starts compounding.
The first time you build something with AI, you may understand almost none of what happened.
That's okay.
Seriously, start there, build it anyway. Then reverse-engineer it. Now you understand a little more. The next time you build something similar, your prompt is better because your thinking is better. You use the right terminology. You understand some of the tradeoffs. You catch a bad decision earlier. You know what questions to ask. You know when the AI is hand-waving. You build something more sophisticated.
Then you take that apart too.
Now you've learned more.
Which means the next thing starts from an even higher floor.
So the loop becomes:
Build. > Dissect. > Understand. > Internalize. > Build something harder.
Then repeat.
That's the flywheel. And once it starts spinning, something very interesting happens. AI stops merely increasing what you can produce. It starts increasing what you can understand.
THIS IS WHY "AI MAKES YOU DUMB" IS BOTH RIGHT AND WRONG
You've probably heard some version of this already.
AI is going to make people stupid.
And yeah.
It absolutely can.
If the loop is: Ask. > Receive. > Paste. > Move on.
Do that for long enough and you're training yourself to become a very efficient endpoint for somebody else's cognition.
The machine knows. The machine writes. The machine decides. The machine explains.
You consume.
That's not an especially compelling vision of the future. And if enough of us do it, we could end up in this very strange civilization where people possess extraordinary capabilities while understanding less and less about the systems producing them.
That's entirely possible. But it isn't inevitable.
Because there's another loop.
The machine gets you somewhere you couldn't have reached efficiently on your own. Then you study the destination. You take apart the vehicle. You learn the principles. You fold some of them into your own thinking. And the next time, you arrive with more of yourself intact.
AI can make intellectual laziness incredibly cheap. It can also make curiosity incredibly productive. Both are true.
The difference is what you do after the answer appears.
GROK THE DAMN THING
There's a word I keep coming back to here.
Grok.
Yeah, there's now an AI called Grok. That's not why the word matters. Robert Heinlein used it in Stranger in a Strange Land to describe something deeper than ordinary understanding.
To grok something is to understand it so completely that it becomes part of your way of seeing it.
That's what I'm after.
You don't have to become an expert in everything AI helps you build. That's impossible. And frankly, it would destroy the advantage. I don't need to become an embedded systems engineer because I built a sensor. I don't need to become a career software developer because I built an application.
But I should know more afterward than I knew before. I should understand the architecture. I should understand the major pieces. I should know where the failure points are. I should know what I'd change next time. I should be able to look at a similar system and recognize some of its anatomy. I should walk away carrying something.
Grok what you build.
THE THING ISN'T THE PRIZE
This may be the part I care about most.
AI makes it incredibly easy to confuse the thing you created with the reward.
Look...I made an app.
Look...I made a device.
Look...I built a website.
Look...I automated this process.
Great.
Those things matter.
Ship them. Use them. Sell them. Be proud of them. But there's another product being built at the same time.
You.
Did the project leave anything behind in your head? Did you acquire a concept? Did you understand another system? Did you pick up vocabulary? Did you learn where one discipline touches another one? Can you now see something you couldn't see before?
Because if the answer's yes, you've acquired something far more durable than the artifact.
You've acquired cognitive capital. And that stacks.
THIS GETS BIG VERY QUICKLY
This isn't really about programming. Or ESP32s. Or websites.
Those are just easy examples. The same thing applies to research.
Finance.
Law.
Biology.
History.
Design.
Engineering.
Infrastructure.
Almost anything.
AI can get you much farther into a domain before your current expertise runs out. That's the first miracle. The second is that it can then turn around and help you understand the domain you just entered.
Historically, the time cost alone put enormous limits on how many fields one person could seriously explore. You could specialize deeply. Or sample broadly. Doing both was expensive.
AI changes some of that economics. It can compress the mechanical work enough that curiosity gets room to breathe.
That's an extraordinary opportunity.
But only if we take it.
CRASH MORE UFOS
So maybe my old alien metaphor wasn't wrong. It was just early.
Four years ago, I was telling people to imagine that something unimaginably advanced was headed toward Earth. Something that would force us to reconsider what humans were capable of.
Now I think we've reached the next scene. The spacecraft is here.
Actually, that's not quite right.
There are spacecraft falling out of the sky every day now.
A new model...A new agent...A new tool...A new capability.
And increasingly, we can drag pieces of that technology into our own lives and make things with it that would've looked ridiculous a few years ago.
Good. Do that.
Crash one in your backyard. Make something amazing with it. Then get the screwdriver. Take the panels off. Trace the wires. Ask stupid questions. Ask better questions. Break things. Put them back together. Understand a little more than you did yesterday.
Then use that understanding to build something even stranger tomorrow.
That's the loop I want. Not artificial intelligence endlessly producing things for me while I slowly become less necessary to the process.
Artificial intelligence expanding the surface area of what I can reach, while curiosity expands the amount of it I actually understand.
The UFO isn't the prize.
Knowing how it flies is.
Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.
Comments