(sparkles trilling) - Magic is creating the illusion that you can defy the laws of nature.

(magical music) (upbeat music) ♪ Hey ♪ (upbeat music continues) ♪ Hey ♪ (upbeat music) (audience laughs) (upbeat music) (audience laughs) (upbeat music continues) (magical music) - [Audience] Whoa!

- [Audience Member] How did he do that?

(audience cheers) - Wow!

- Did you see that?

- [Announcer] Magic is creating the illusion that you can defy the laws of nature.

- Science is the study of how the natural world works.

This is "Science."

(audience cheers) (drone buzzes) Whoa!

Artificial intelligence.

Sounds cool, but what exactly is it?

Let's break it down.

Artificial means fake, not real, or in this case, not made by humans.

Intelligence means the capacity for learning or for reasoning.

So it's learning not done by humans.

So when we say artificial intelligence do we mean fake brains?

- [Pine Cone] It's alive!

- No, not really.

Artificial intelligence, or AI, is intelligence demonstrated by a machine rather than a living thing.

Some computers can be programmed to learn things on their own.

Of course, this doesn't mean they learn like humans learn.

Instead, humans program computers with codes that tell them what to do under certain conditions.

Those computers can then apply parts of these codes, called functions, to solve specific problems.

Computers that solve problems are said to have artificial intelligence.

Whoa!

(drone buzzing) I wish my quadcopter drone had a brain.

Even a fake one would be good.

Right now, I'm trying to learn to fly this quadcopter (chuckles) and I'm pretty bad at it.

Okay, let's see.

(drone buzzing) If the quadcopter is flying towards me, I press this lever.

If it's not flying towards me, I press it that way.

Did you see what I did?

I made a decision based on the condition of the quadcopter's movement towards me.

My brain made a Boolean decision.

A Boolean decision means I assessed the situation and made a decision based on whether a condition was true or false.

We use Boolean statements when programming computers to make choices just like I used my brain.

(drone whizzes) Whoa!

Obviously not good yet.

We humans have amazing brains and we can learn stuff on our own.

So you may ask why do we need artificial intelligence?

For one thing, dangerous or dirty jobs that people don't want to do can be automated and AI can help us do those jobs.

Here is an example.

I do not like the vacuum.

It's not dangerous, but it's dirty.

Hey, maybe you love to vacuum, but not me.

It is super boring.

So I bought an automatic vacuum cleaner A robotic vacuum cleaner.

An AI vacuum cleaner.

♪ I don't have to vacuum ♪ ♪ I don't have to vacuum ♪ ♪ I don't have to vacuum ♪ Once a day, my automatic vacuum, or smart vacuum, rolls through my house and vacuums the floors.

I love it.

It does it all by itself.

(smart vacuum humming) (vacuum honks) Oh, excuse me.

There it goes now doing its thing and I don't have to vacuum.

It has learned the shape of my home and it vacuums all over my house.

It vacuums everywhere.

Thank you, AI vacuum.

(vacuum honks) When it's finished vacuuming, it drives itself back to a charger to get ready to do it again tomorrow.

Some people don't like robotic vacuum cleaners and insist humans do a better job.

What do you think?

My brother doesn't like his robotic vacuum cleaner because the color doesn't match his favorite rock.

Is that a logical argument against the vacuum cleaner's ability to clean my floor?

I don't think so.

Artificial intelligence isn't just for simple stuff that we don't like to do, like vacuuming the floor, it can also sort through huge amounts of data to do very complicated stuff like predicting the weather.

Sure, humans program computers to do these things, but computers don't get tired or distracted like humans do, and they can multitask.

This makes artificial intelligence more efficient than human intelligence in certain situations.

As AI gets better and better, there's almost no limit to what it can do for us.

(drone buzzing) Whoa!

(dramatic music) The AI Big Five.

We are going to examine five big ideas in artificial intelligence.

Number one, how do computers perceive things?

Computers can use many different types of sensors to gather information about their surroundings.

For example, a computer uses a camera to gather visual information even in the dark.

Look at all of those cool colors my infrared camera detects.

It can detect things that my own eyes can't see.

Groovy, man.

There's even a camera on my quadcopter.

(drone buzzing) Whoa, man, that was really close.

The camera on my quadcopter is not hooked into computer sensors.

It is definitely not analyzing any data to not whack me in the head.

If it had AI, it might have recognized that it was about to whack me in the head.

It might have sensed that I was standing here and avoided hitting me.

Using AI it could do this by itself.

I kind of, I kind of wish it could do that.

Computers use sensors in the way that we use our eyes, ears, hands, and even our noses and tongues.

What kind of sensors could AI use besides cameras?

How about microphones like this one?

Hello?

How might we use a microphone?

Microphones sense sound.

Maybe you'd put one in a hot, steamy, dangerous jungle instead of going there yourself to identify the songs of birds that live there.

Or maybe you could even put it underwater to listen to critters in the deepest parts of the ocean.

Are fish singing songs at the bottom of the Mariana Trench, which is seven miles beneath the ocean surface?

♪ Give my regards to Broadway ♪ ♪ Remember me to Herald Square ♪ ♪ Tell all the gang at 42nd Street ♪ ♪ That I will soon be there ♪ (fish gurgles) - I knew it.

Those fish can sing.

Man, they are really good.

I bet I could get them a lead role in the Broadway show "Cats."

(cat meows) How could we use information about singing fish?

Hmm, what problem would it solve and how could it benefit humankind?

Hmm.

Maybe I could use those fish on my show.

Send me your resume.

(light upbeat music) The ability to perceive conditions around them is the first step for computers in being able to make decisions.

The second big idea is that computers need data representations of the world they're going to work within.

Data representations are a type of information map.

Computers compare the information they gather from their sensors against the map and construct new data maps based on the rules of the old data map.

Is that confusing?

Let me see if I can help.

In my head I know what my neighborhood looks like.

You know what yours looks like.

When I need to go somewhere, I think of the directions and I begin my trip.

I've compared the picture in my head with the data I'm getting from my eyes.

Remember, eyes are sensors sending data to our brains.

I'm comparing what I'm seeing with the image, or map, of my neighborhood in my head.

If they match, everything is going according to plan.

If there is something different, I'll need to adjust and learn.

If a box falls off a truck and into the road, I'm going to have to adjust my route.

What would I do next?

I'd stop in the road at the box.

I know I have to avoid the box and not crash into a car or the side of the road.

There goes my quadcopter.

I better catch that thing soon.

Just like the map in my head A computer has a type of information map against which it compares what it is sensing.

If I were in a car with AI, its computer sensors would discover something in the road and it would stop just like I did.

That moves us to the next big idea.

Number three, computers can learn from data.

The computer senses that something is in the road and to get to its destination, it has to solve this unexpected problem.

Something's in the way.

When the computer compares its own information map with the newly gathered information, and it follows its programmed function to make a decision, it stores that decision in its memory.

It's pretty simple for a human to decide to drive around the box, but for a computer, man, that is difficult.

The computer has to have a huge amount of information to make decisions like this, and it must use a sensor, like a camera, to see if a car is coming and compute how much time it has to get around the object without flipping over or screeching its tires.

There are a million things that can go wrong and they all have to be programmed into the computer.

The AI computer must take these factors into account.

Is the road icy?

Are there a lot of people on bicycles?

Is there a squirrel about to run across the road?

Will a box jump or move?

Humans sense all of these things quickly and easily.

We do this almost automatically without thinking because we've had a lifetime of learning and assessing challenges and because our brains are amazing.

A computer can be programmed to perceive, analyze, and store all the data about every situation it encounters.

Every time a computer does this, it's memory improves.

Let's take a quick break for a little magic.

We'll be back with part two of The Five Big Ideas of AI in just a minute.

(sparkles jingle) Magic is creating the illusion that you can defy the laws of nature.

(whimsical music) (upbeat music) ♪ Hey ♪ (upbeat music) (audience laughs) (upbeat music) ♪ Hey ♪ (upbeat music) (audience laughs) (upbeat music) - [Audience Member] Whoa.

(audience chattering) (upbeat music) (audience cheers) - Man, we are lucky.

We get to live in a world with computers, tablets, cell phones, and all kinds of amazing digital devices.

We can connect with our friends, our families, and people all over the world through screens.

I call this the Pretend Magical World.

It's pretend because a screen is between you and the world.

We also get to live in a world where we connect with our friends, our families, people around the world, and with nature, with no screens.

I call this the Real Magical World.

It's real because there's no screen between us.

To be happy, healthy, 21st century citizens, we should spend as much time in the Real Magical World with no screens, playing, talking, hanging out, having fun as we do in the Pretend Magical World.

It makes us better people.

We are lucky to have both magical worlds.

Use them wisely.

Artificial intelligence.

Let's review the Five Big Ideas one through three.

Big idea number one, computers use sensors to perceive things.

Big idea number two, computers need data representations of the world they're going to work within.

Big idea number three, computers can learn from data.

That brings us to big idea number four, natural interactions.

Oh boy, this is a tough one.

During a natural conversation I might say to you, "Did you see that?

How's your brother's rock?"

Or, "My shoe came off," and it would make perfect sense to you.

We humans use many cues outside of language to communicate.

We often will refer to previous conversations, events, or even jokes.

Have you ever met someone who had an accent that you could barely understand?

Even though you are both speaking the same language you may have had trouble communicating with them.

There are so many variations in how people speak that machines sometimes struggle to understand us.

Not only are different languages complicated, but even accents within a common language can be confusing.

People that speak the same language can have different accents and pronounce words differently.

For instance, I live in the southern United States.

- [Southerner] Hey, y'all.

- I could say, "I love to recycle aluminum cans."

But my friend Colin in England, or the UK, might say the same thing this way.

- [Colin] I love to recycle aluminum cans.

- See, accents are fairly easy for humans to comprehend but they can sometimes be a challenge for machines.

(cell phone buzzes) Hold on a second.

I'm sorry, this is really rude, but I've got to do this now.

I'll see you for lunch.

Cool, right?

So when you're in a hurry you can use voice to text.

(phone buzzes) Aw, man.

So sorry.

No, no, I-I-I didn't say, "I think you're Captain Crunch."

I'll see you for lunch.

Now where were we?

(phone buzzes) No, no, I-I don't want to give you a punch.

Oh, man.

(phone ringing) Excuse me.

Hello.

(caller mumbles) No, no, no, no, I said, "I'll see you for lunch."

Yes, I'll see you for lunch.

Whew, that was confusing.

AI systems are getting better as they learn to use language but right now it's not really very natural.

It's still learning and improving.

It's one of the big ideas AI is trying to solve.

But the possibilities are truly amazing.

Here's the thing, human communication is complex, but every time you have to repeat yourself, the computer adds that experience to its memory and gets better.

In other words, it improves the validity and accuracy of its dataset based on social interactions.

Did you get that?

Every time you have to repeat yourself, the computer adds that experience to its memory and gets better.

It's improving the validity and accuracy of its dataset based on social interactions.

Don't forget, we humans can do some pretty amazing things.

AI is not a substitute for people.

It's a tool that we can use to help us do things better and safer.

AI, artificial intelligence, can do so many more calculations than we can, and it can do them much more quickly.

Let's say you have a favorite video game.

It's fun and you love to play it.

You play it as often as possible.

Each time you play, you learn from your mistakes, and you get better the next time you play.

If you play this game 10 times, you'll get better.

If you play this game a hundred times, you'll get a lot better.

If you play this game a thousand times, you'll be a ton better.

You have learned from your mistakes each time you played, but AI, in some cases, can play the whole game in just seconds.

It took you years to get a lot better at that game.

It took AI a few minutes to get a lot better at that game.

This is useful to us humans because we have trouble analyzing huge sets of data and AI and computers don't.

AI computers can evaluate millions of bits of information in seconds.

Here's a really good example of using huge amounts of data to create something really cool.

DALL-E and DALL-E 2 are machine learning models, or AI, that we're developed by the company OpenAI.

This cool software creates digital images from natural language descriptions.

I gave it some pretty complicated instructions.

I said, "Make me an image of a tornado destroying a peanut butter and jelly sandwich."

This is what it came up with.

Notice that the peanut butter is in the shape of a tornado and it's inside a sandwich with jelly, so it's pretty close.

It's not perfect, but what does a peanut butter and jelly sandwich being destroyed by a tornado actually look like?

I have no idea, so I think it's pretty good.

Next, I told it to give me a cute hamster destroying a peanut butter and jelly sandwich.

Apparently I was in a destroying kind of mood that day.

These are pretty good.

They kind of do look like a hamster destroying a peanut butter and jelly sandwich.

Most important of all, the AI returned multiple image choices for me to choose from.

It even decided that I might want a cartoon version of a hamster destroying a peanut butter and jelly sandwich.

So DALL-E 2 AI also gave that to me as a choice.

Remember, AI can choose from a very, very large set of data.

DALL-E 2 even made the decision to give me multiple image versions as a photorealistic image.

If I wanted to use a cartoon version, it gave me that choice too.

Lastly, I asked it for a happy man with a giant peanut butter jelly sandwich.

I wasn't sure what I would get, and this is what I got.

Okay, that's super weird.

So there's still a lot of work to be done, but wow, AI is really exciting.

Now we're onto the Big Idea Number Five.

- [Audience Member] Oh yeah.

- What will the changes to society be?

That's a huge question and one that your generation will be grappling with for a long time.

Technology impacts human society in many unexpected ways.

It always has.

The person who invented the automobile probably never imagined drive-through dining, or smog check stations, or RV parks, or the need for speed bumps.

So what will AI bring?

We are already using many assistive technologies, like phones that recognize our voices and follow our commands.

And when you ask voice activated technology to play music in your home, you are using artificial intelligence.

How do you think those technologies have helped us humans?

We know travel, work, communication, security, space exploration, and even healthcare will see big changes in the years to come, thanks to AI.

So let's try something.

Think of a problem you have to solve regularly.

Maybe you can't wake up in the morning for breakfast, or you have a hard time making friends, or you might even think about world problems that need to be solved, like there's too much plastic pollution in the world's oceans, or batteries for electric cars are too expensive, or make too much pollution being made, or are slow to charge.

Can you imagine how artificial intelligence might help you in the future?

Think of the wildest solutions possible.

That's how engineers come up with solutions to problems.

They brainstorm solutions.

Now, try to make as logical an argument as you can as to why AI would be a good solve for this problem.

Now, let's switch to the other side of creating solutions.

Let's think about the problems.

Think of all the ways your new smart computer could go wrong.

This is part of the problem solving process as well.

You must consider all the solutions and the problems when deciding how to use AI.

You are the next generation of creators and problem solvers that will be writing the code that does so many amazing things.

I can't wait to see how you use artificial intelligence to make the world better.

There are so many problems in the world now and we all must use our hearts, minds, and creative vision to make it better.

None of us knows what the future will bring but we do know this, it's going to be really interesting.

(drone buzzing) (tires screech) (drone shatters) Oh no, oh!

Ah, man!

Huh, not to worry, the automatic vacuum will clean that up.

(vacuum honks and whirs) (magical trilling) - And now something silly.

(upbeat music) - (laughs) She said she didn't work nearly as hard as I did to get pretty.

Noses and tongues.

(upbeat music) People that speak in the same language can have different accents and I don't remember the rest of that line.

So let's break it into two.

(laughs) I am monologuing.

That's my line?

(crew murmuring) As hard as I did to get pretty.

She just threw me under the bus.

(grunts) Did you feel that?

- [Director] Last line was "Noses and tongues."

(upbeat music) - [Singer] Biologically, chemically, atomically, everything is connected.

(upbeat music) ♪ Everything is connected ♪ ♪ Everything ♪ - [Announcer] Major funding for this program provided by Jason Isbell and Amanda Shires, Keep Alabama Beautiful, and Northrop Grumman.

♪ Yes, the ocean is deep ♪ ♪ And the sky is so high ♪ ♪ And the Earth is so wide ♪ - [Singer] It's a mystery.

But deep down everyone knows.

(upbeat music) ♪ Everything is ♪