Learning

Podcast: How We Taught Computers to Think

May 2026  ·  9 min read

Why I Created This Podcast

I built this podcast for my kids, but I think it’s helpful for everyone who wants to understand how AI actually works—not as magic or something to fear, but as a tool with a real history and clear limitations.

The podcast was generated using Google NotebookLM, which transformed research materials into a conversational dialogue. It’s a good example of AI as a teammate in education—amplifying intent and curation, not replacing it.

Full sources and links are at the bottom of this post.

Listen to the Podcast

Full Transcript

Have you ever asked a smart speaker in your kitchen to play your absolute favorite song? Or maybe you’ve wondered how your streaming app knows exactly which cartoon to suggest to you next. Yeah. It almost feels like magic, right? But it’s actually artificial intelligence or AI.

Exactly. And today we are going on a huge time traveling adventure for this deep dive. We’re going to discover exactly how people taught computers to actually think.

We really are. We’ve brought a giant stack of fun history books, kid-friendly coding guides, and just some amazing stories about crazy inventors. We’re going to break all of this down so simply that by the end of this deep dive, you will be an absolute AI expert.

You really will. We’ll learn how AI is just like a baby learning to spot a dog. And this is my favorite part, a giant stack of pancakes.

I am so ready for the pancakes. But before we can understand the AI living inside your house right now, we have to travel back in time.

Way way back. Thousands of years ago, long before electricity was even invented, just to see where the whole idea of a thinking machine even came from.

It’s crazy, but the dream of AI is actually ancient. In Greek mythology, there were stories about this guy named Talos.

Wait, who was Talos?

He was a giant bronze man who guarded an island, a mythological robot, basically.

Whoa. A giant bronze robot. That is super cool.

Yeah. And then around the year 1200, an inventor named Al Jazari actually built a real boat that had a robot band on it.

A robot band like the ones at a pizza arcade.

Exactly like that. They played real music to entertain people. People have always wanted to build machines that act like us.

Okay. So, let’s fast forward a bit to the 1800s. There’s this inventor Charles Babbage and he designed something called the analytical engine.

This thing was wild. It was a giant steam powered computer.

Wait, wait, wait. Steam powered like a choo choo train.

Exactly. Imagine trying to build a modern smartphone but out of heavy metal gears, clunky wheels and hissing steam.

That sounds incredibly heavy and loud.

It was. But his friend Ada Lovelace, she looked at his blueprints and realized something amazing.

What did she see?

She realized this giant steam machine could do way more than just math. She figured out it could create music or even art.

No way.

Yes. She basically wrote the very first computer program before real computers were even built.

That is so smart. And around that same time, another guy named George Boole invented a super simple language for these future computers. And it only used two words.

Those two words, what were they?

True and false. That’s it. He proved that really big, complex human thoughts could be broken down into tiny, simple math steps.

So these early dreamers were laying down the blueprints, but it took a massive world war and a brilliant hero to actually build the first real thinking machines.

Yeah, those ideas were just on paper until World War II. That’s when we meet Alan Turing. I love his story. He was a boy who could solve super advanced math totally in his head.

He was a genius and his best friend Christopher really inspired him to dream big.

So during the war, Turing went to work at this highly secret place called Bletchley Park.

Super secret. The German Navy had this unbreakable secret code called Enigma.

And Turing helped crack it. He built a special machine called the Bombe.

Not a bomb that explodes, right? Just to be clear.

Good point. It was spelled B-O-M-B-E. It was an electromechanical machine that tested thousands and thousands of settings to read those secret messages.

He essentially built a machine to do the heavy thinking for them. And after the war in 1950, Turing asked a really massive question.

Can machines think?

Exactly. But thinking is hard to measure. So he invented a game to test it called the Imitation Game.

People call that the Turing Test now, right?

They do. The rules are simple. If a computer can talk to you in a chat room and it fools you into thinking it’s a real live human being, then it passes the test.

You got it. It proves it has artificial intelligence.

Wait, hold on though. Did Turing actually build a robot that could pass this test back in 1950?

Oh, no. Definitely not. The technology wasn’t there yet. He was just drawing the ultimate finish line for the future.

Ah, okay. So he wasn’t asking what is a thought. He was just asking can a machine imitate us.

Exactly. And that totally changed the rules of the game for all the scientists who came after him.

And because of Turing’s big challenge, scientists in the 1950s got super excited. They were like, “We are going to build a thinking machine right now.”

They were so excited. In 1956, a bunch of scientists had this giant summer camp at Dartmouth College.

A science summer camp sounds amazing, right?

And that is where they officially invented the name artificial intelligence.

And they had some early wins, too. They built a computer program named ELIZA that pretended to be a therapist.

Yeah. You could type to ELIZA about your feelings, and it would ask you a few questions back.

And they built Shaky. Shaky was this really cute robot that could look at a room and push blocks around.

But then they kind of hit a giant brick wall.

Oh no. What happened?

Well, the scientists tried to teach these computers using giant rule books. They called it symbolic AI.

When you say giant rulebooks, how big are we talking?

Humongous. There was one program called XCON and it had 50,000 rules.

50,000 rules. That is impossible to remember.

And that was the problem. If you change just one single rule, the whole system just completely broke down.

It’s kind of like trying to build a giant Lego castle, but you only have 100 blocks.

Or trying to teach a kid how to ride a bike by making them sit down and read a 10,000 page instruction manual.

Oh my gosh, that would be the worst way to learn how to ride a bike. You just have to get on it and pedal.

Exactly. The scientists realized you can’t just program common sense. You can’t write a math rule that explains water is wet. It’s just too complicated. So because it was too hard, the money ran out.

And AI had to take a really long sleepy nap. They actually called it the AI Winter.

The AI Winter. That sounds so sad.

It sounds sad, but it was actually a really good thing. It forced the scientists to wake up and realize that simply telling a computer the rules doesn’t work. They needed to let the computer learn the rules all by itself.

Yes. Which leads us out of the AI Winter and into something brand new called machine learning.

Machine learning. This is where they let the computer practice, right? Just like you do when you learn a brand new game.

Exactly. Instead of handing the computer a giant rule book about what a cat looks like, they tried something totally different.

What did they do?

They just showed the computer 10,000 pictures of cats and basically said, “Here, you figure it out.”

Just staring at cat pictures until it learns. That’s hilarious. But how does a computer actually look at a picture?

To do this, they had to invent something called neural networks. It’s basically a pretend brain made out of math.

A pretend brain made of math. That sounds complicated.

It does, but imagine it like a giant stack of pancakes.

Okay. Yes, I was promised pancakes. Let’s hear it.

Okay. So, the first layer of pancakes at the very bottom, those are the lookers.

The lookers. Got it. What do they look for?

They just look for tiny details in the picture like pointy ears or a long furry tail.

Okay. And then what?

Then they pass notes up to the middle layers of pancakes. Those are the thinkers.

Oh, so the thinkers put the pieces together like pointy ears plus tail.

Exactly. And finally, the very top layer of pancakes are the deciders. They take a vote and decide, “Hey, I think this is a cat” or “Nope, it’s a dog.”

Okay, but what if the top pancake guesses wrong? What if it calls a cat a dog?

That’s the best part. If it guesses wrong, a boss robot tells all the little helpers in the pancakes to adjust their tiny dials and switches, like, “Turn down the dog dial. Turn up the cat dial.”

You got it. And they do this thousands of times. So, the more mistakes they fix, the smarter they get.

Exactly.

It’s literally just like practicing soccer or playing a video game. Every time you mess up, you learn and get a little bit better.

That is the real magic of machine learning. Humans aren’t telling the computer how to see a cat anymore. The computer is finding the patterns entirely on its own.

Yep. And once computers mastered finding patterns in pictures, they started finding patterns in games and even in our language, which leads directly to the tools that you might use today. AI got so good at patterns, it started beating humans at really hard games.

Oh yeah. In 1997, an AI named Deep Blue actually beat the human world champion at chess.

Chess is so hard.

I know. And later, another AI named AlphaGo played a super complicated game called Go.

Did it win?

It did, but it didn’t just win. It made a totally new, highly creative move that no human player had ever even thought of.

Wow. It learned the game so well. It invented new ways to play. That is amazing.

It really is. But playing games is one thing. Understanding words is way harder, because language is messy. So, how did they fix that?

In 2017, scientists invented something called transformers.

Like the robots in disguise.

Not quite. This was a special math tool that allowed AI to understand how words connect in a big sentence.

Give me an example.

Okay. If I say the cat sat on the mat because it was tired, the transformer helps the AI know that the word “it” means the cat, not the mat.

Because mats don’t get tired.

Exactly. This big invention led straight to ChatGPT.

ChatGPT. Everyone is talking about that, right?

It’s basically a generative pre-trained transformer, but really you can just think of it as a smart word wizard.

A smart word wizard. I like that. But how does it actually know what to say to you?

Have you ever seen the autocomplete on a tablet or phone keyboard where it tries to guess the next word you want to type?

Oh, yeah. I use that all the time. Sometimes it guesses funny things.

Well, ChatGPT works just like that, but on a massive scale. It predicts the next word based on reading trillions of puzzle pieces of words.

Trillions of puzzle pieces.

Yeah. They call them tokens. So the word hamburger isn’t one word to the AI. It’s three tokens: Ham-burg-er.

Okay. Let me put this back on for a second.

Go for it.

If it’s just guessing the next word, like a giant autocomplete, does it actually know what it’s talking about? Is it always right?

That is the biggest secret about AI right there. No, it doesn’t know facts at all.

Wait, really? It doesn’t know facts.

Not at all. It just knows patterns of words. Sometimes it confidently makes things up entirely. Scientists call that a hallucination.

A hallucination. So, it’s basically just faking it.

In a way. It’s not a thinking human inside a box. It’s really just the world’s best super parrot. It’s beautifully mimicking human pattern.

Okay. A giant pattern matching super parrot. That makes so much sense. But that leads me to a really big question.

What’s that?

If AI is this giant super parrot getting smarter every day, what does that mean for you? Are robots going to take over the world?

Oh, definitely not. You do not need to worry about that.

Phew. Okay, good. But why not?

Because the AI we have today is what we call narrow.

Narrow, like a skinny hallway?

Sort of. It means it’s super smart at one very specific job, like playing chess or generating text. But we don’t have general AI.

General AI would be like Wall-E from the movie, right? A robot that can do lots of different things and make decisions.

Exactly. We don’t have Wall-E. And more importantly, AI doesn’t have any feelings, right? It can beat you at a game of chess, but it can’t give you a hug when you’re feeling sad.

Nope. AI is not here to take our place. It’s here to be our teammate.

I love that. A teammate like Batman has Robin or Iron Man has Jarvis.

Yes, humans have AI. It’s a tool to help us be awesome.

So, if AI is going to be our teammate, how do you get ready for the future? How do you build an AI-ready brain? By focusing on the exact human skills that AI completely lacks.

Like what?

Like creativity. AI can’t ask “what if” and critical thinking. AI never stops to ask why. It just does math.

So humans have the logic and the imagination.

Exactly. And the best way to build those skills isn’t by staring at a computer screen all day long.

Here’s where it gets really interesting to me. You don’t even need a computer to learn the skills that built the computer.

Not at all. You build your AI-ready brain by playing with wooden blocks, solving paper mazes, and doing fun logic puzzles.

That is so cool. So, playing in the backyard actually prepares you for the future of technology.

It absolutely does because humans have hearts and curiosity and the ability to ask the right questions. The AI is just the parrot giving you an answer, but a creative human is the one who knows which amazing questions to ask in the first place.

Exactly.

Well, what an incredible time-traveling deep dive this has been.

We covered so much ground today.

We really did. We went all the way from ancient Greek myths about bronze robots to giant steam machines, to pancake brains and super parrots.

It is a crazy history.

It really is. So, we want to leave you with a thought to think about today.

Yeah. Think about this. What is one thing that you would want to teach an AI?

That’s a great question. What pattern or skill would you want a computer to learn from you?

Maybe you teach it how to write really funny knock-knock jokes. Or maybe how to recognize all the different bugs and animals in your backyard. Or maybe you could teach it how to be kinder to people.

I love that. Think about it because you might just be the person who inspires the next big AI breakthrough. You never know.

Thank you so much for joining us on this adventure. Goodbye and keep asking amazing questions.

Sources

Podcast created with Google NotebookLM — transforming research materials into conversational audio.