0:00:00.080,0:00:02.120
Hey there! I’m Jabril.

0:00:02.120,0:00:04.759
John Green Bot: And I am John Green Bot

0:00:04.760,0:00:08.160
and welcome to Crash Course Artificial Intelligence.

0:00:08.160,0:00:10.780
Now, I want to make sure we’re starting[br]on the same page.

0:00:10.780,0:00:12.740
Artificial intelligence is everywhere.

0:00:12.750,0:00:17.180
It’s helping banks make loan decisions,[br]and helping doctors diagnose patients, it’s

0:00:17.180,0:00:21.450
on our cell phones, autocompleting texts,[br]it’s the algorithm recommending YouTube

0:00:21.450,0:00:23.199
videos to watch after this one!

0:00:23.199,0:00:26.710
AI already has a pretty huge impact on all[br]of our lives.

0:00:26.710,0:00:30.610
So people, understandably, have some polarized[br]feelings about it.

0:00:30.610,0:00:35.410
Some of us imagine that AI will change the[br]world in positive ways, it could end car accidents

0:00:35.410,0:00:40.120
because we have self-driving cars, or it could[br]give the elderly great, personalized care.

0:00:40.120,0:00:44.080
Others worry that AI will lead to constant[br]surveillance by a Big Brother government.

0:00:44.080,0:00:46.310
Some say that automation will take all our[br]jobs.

0:00:46.310,0:00:50.200
Or the robots might try and kill us all.

0:00:50.200,0:00:52.380
No, we’re not worried about you John Green Bot.

0:00:52.380,0:00:55.860
But when we interact with AI that’s currently[br]available like Siri...

0:00:55.860,0:00:58.020
Hey Siri.

0:00:58.020,0:01:00.300
Is AI going to kill us all?”

0:01:00.300,0:01:03.500
Siri: “I don’t understand ‘Is AI going[br]to kill us all.’”

0:01:03.500,0:01:06.680
… it’s clear that those are still distant[br]futures.

0:01:06.680,0:01:11.940
Now to understand where artificial intelligence might be headed, and our role in the AI revolution,

0:01:11.940,0:01:15.340
we have to understand how we got[br]to where we are today.

0:01:15.340,0:01:24.340
[INTRO]

0:01:24.340,0:01:29.120
If you know about artificial intelligence[br]mostly from movies or books, AI probably seems

0:01:29.120,0:01:32.640
like this vague label for any machine that[br]can think like a human.

0:01:32.640,0:01:37.500
Fiction writers like to imagine a more generalized[br]AI, one that can answer any question we might

0:01:37.500,0:01:39.760
have, and do anything a human can do.

0:01:39.760,0:01:45.800
But that’s a pretty rigid way to think about[br]AI and it’s not super realistic.

0:01:45.800,0:01:49.880
Sorry John Green-bot, you can’t do all that yet.

0:01:49.880,0:01:54.880
A machine is said to have artificial intelligence[br]if it can interpret data, potentially learn

0:01:54.880,0:01:59.540
from the data, and use that knowledge to adapt[br]and achieve specific goals.

0:01:59.540,0:02:04.480
Now, the idea of “learning from the data”[br]is kind of a new approach.

0:02:04.480,0:02:05.950
But we’ll get into that more in episode[br]4.

0:02:05.960,0:02:10.480
So let’s say we load up a new program in[br]John Green-bot.

0:02:18.280,0:02:22.950
This program looks at a bunch of photos, some[br]of me and some of not of me, and then learns

0:02:22.950,0:02:24.120
from those data.

0:02:24.120,0:02:28.170
Then, we can show him a new photo, like this[br]selfie of me here in the studio filming this

0:02:28.170,0:02:32.360
Crash Course video, and we’ll see if he[br]can recognize that the photo is me.

0:02:32.360,0:02:38.160
John Green Bot: You are Jabril.

0:02:38.160,0:02:42.340
If he can correctly classify that new photo,[br]we could say that John Green-bot has some

0:02:42.340,0:02:43.970
artificial intelligence!

0:02:43.970,0:02:49.270
Of course, that’s a very specific input[br]of photos, and a very specific task of classifying

0:02:49.270,0:02:53.120
a photo that’s either me or not me.

0:02:53.120,0:02:57.390
With just that program John Green-bot can’t[br]recognize or name anyone who /isn’t/ me…

0:02:57.390,0:03:01.450
John Green Bot: You are not Jabril.

0:03:01.450,0:03:12.980
He can’t navigate to places.

0:03:12.980,0:03:16.880
Or hold a meaningful conversation.

0:03:16.880,0:03:18.400
No.

0:03:18.400,0:03:20.440
I just don’t get it.

0:03:20.440,0:03:26.440
Why would anyone choose a bagel when you have[br]a perfectly good donut right here?

0:03:26.440,0:03:32.440
John Green Bot: You are Jabril

0:03:32.440,0:03:39.120
Thanks John Green Bot.

0:03:39.160,0:03:44.440
He can’t do most things that humans do,[br]which is pretty standard for AI these days.

0:03:44.440,0:03:49.000
But even with this much more limited definition of artificial intelligence, AI still plays

0:03:49.000,0:03:51.520
a huge role in our everyday lives.

0:03:51.520,0:03:57.569
There are some more obvious uses of AI, like Alexa or Roomba, which is kind of like the

0:03:57.569,0:03:59.890
AI from science fiction I guess.

0:03:59.890,0:04:02.660
But there are a ton of less obvious examples!

0:04:02.660,0:04:07.790
When we buy something in a big store or online,[br]we have one type of AI deciding which and

0:04:07.790,0:04:09.950
how many items to stock.

0:04:09.950,0:04:14.680
And as we scroll through Instagram, a different[br]type of AI picks ads to show us.

0:04:14.680,0:04:20.809
AI helps determine how expensive our car insurance[br]is, or whether we get approved for a loan.

0:04:20.809,0:04:23.479
And AI even affects big life decisions.

0:04:23.479,0:04:28.099
Like when you submit your college (or job)[br]application AI might be screening it before

0:04:28.099,0:04:29.300
a human even sees it.

0:04:29.300,0:04:32.990
The way AI and automation is changing everything,[br]from commerce to jobs, is sort of like the

0:04:32.990,0:04:35.830
Industrial Revolution in the 18th century.

0:04:35.830,0:04:41.550
This change is global, some people are excited[br]about it, and others are afraid of it.

0:04:41.550,0:04:46.159
But either way, we all have the responsibility[br]to understand AI and figure out what role

0:04:46.159,0:04:48.229
AI will play in our lives.

0:04:48.240,0:04:51.439
The AI revolution itself isn’t even that[br]old.

0:04:51.440,0:04:55.360
The term artificial intelligence didn’t[br]even exist a century ago.

0:04:55.360,0:04:59.200
It was coined in 1956 by a computer scientist[br]named John McCarthy.

0:04:59.210,0:05:03.759
He used it to name the “Dartmouth Summer[br]Research Project on Artificial Intelligence.”

0:05:03.759,0:05:06.150
Most people call it the “Dartmouth Conference”[br]for short.

0:05:06.150,0:05:10.289
Now, this was way more than a weekend where[br]you listen to a few talks, and maybe go to

0:05:10.289,0:05:11.740
a networking dinner.

0:05:11.740,0:05:15.189
Back in the day, academics just got together[br]to think for a while.

0:05:15.189,0:05:19.729
The Dartmouth Conference lasted eight weeks[br]and got a bunch of computer scientists, cognitive

0:05:19.729,0:05:22.779
psychologists, and mathematicians to join[br]forces.

0:05:22.779,0:05:26.689
Many of the concepts that we’ll talk about[br]in Crash Course AI, like artificial neural

0:05:26.689,0:05:31.009
networks, were dreamed up and developed during[br]this conference and in the few years that

0:05:31.009,0:05:32.009
followed.

0:05:32.009,0:05:36.809
But because these excited academics were really[br]optimistic about artificial intelligence,

0:05:36.809,0:05:38.389
they may have oversold it a bit.

0:05:38.389,0:05:43.419
For example, Marvin Minsky was a talented[br]cognitive scientist who was part of the Dartmouth

0:05:43.419,0:05:44.499
Conference.

0:05:44.499,0:05:49.689
But he also had some ridiculously wrong predictions[br]about technology, and specifically AI.

0:05:49.689,0:05:54.449
In 1970, he claimed that in "three to eight[br]years we will have a machine with the general

0:05:54.449,0:05:57.000
intelligence of an average human being."

0:05:57.000,0:05:58.729
And, uh, sorry Marvin.

0:05:58.729,0:06:01.289
We’re not even close to that now.

0:06:01.289,0:06:05.059
Scientists at the Dartmouth Conference seriously[br]underestimated how much data and computing

0:06:05.059,0:06:08.680
power an AI would need to solve complex, real[br]world problems.

0:06:08.680,0:06:13.729
See, an artificial intelligence doesn’t[br]really “know” anything when it’s first

0:06:13.729,0:06:15.479
created, kind of like a human baby.

0:06:15.479,0:06:20.319
Babies use their senses to perceive the world[br]and their bodies to interact with it, and

0:06:20.319,0:06:23.099
they learn from the consequences of their[br]actions.

0:06:23.099,0:06:27.350
My baby niece might put a strawberry in her[br]mouth and decide that it’s tasty.

0:06:27.350,0:06:30.889
And then she might put play-doh in her mouth[br]and decide that it’s gross.

0:06:30.889,0:06:36.379
Babies experience millions of these data-gathering[br]events as they learn to speak, walk, think,

0:06:36.379,0:06:37.470
and not eat play-doh.

0:06:37.470,0:06:42.080
Now, most kinds of artificial intelligence[br]don’t have things like senses, a body, or

0:06:42.080,0:06:46.729
a brain that can automatically judge a lot[br]of different things like a human baby does.

0:06:46.729,0:06:49.699
Modern AI systems are just programs in machines.

0:06:49.699,0:06:52.599
So we need to give AI a lot of data.

0:06:52.599,0:06:57.550
Plus, we have to label the data with whatever[br]information the AI is trying to learn, like

0:06:57.550,0:06:59.869
whether food tastes good to humans.

0:06:59.869,0:07:04.309
And then, the AI needs a powerful enough computer[br]to make sense of all the data.

0:07:04.320,0:07:07.120
All of this just wasn’t available in 1956.

0:07:07.120,0:07:12.569
Back then, an AI could maybe tell the difference[br]between a triangle and a circle, but it definitely

0:07:12.569,0:07:16.039
couldn’t recognize my face in a photo like[br]John Green-bot did earlier!

0:07:16.039,0:07:22.099
So until about 2010 or so, the field was basically[br]frozen in what’s called the AI Winter.

0:07:22.099,0:07:27.690
Still there were a lot of changes in the last[br]half a century that led us to the AI Revolution.

0:07:27.690,0:07:32.399
As a friend once said: “History reminds[br]us that revolutions are not so much events

0:07:32.399,0:07:33.960
as they are processes.”

0:07:33.960,0:07:38.880
The AI Revolution didn’t begin with a single[br]event, idea, or invention.

0:07:38.880,0:07:43.379
We got to where we are today because of lots[br]of small decisions, and two big developments

0:07:43.379,0:07:44.400
in computing.

0:07:44.400,0:07:48.830
The first development was a huge increase[br]in computing power and how fast computers

0:07:48.840,0:07:50.200
could process data.

0:07:50.200,0:07:53.760
To see just how huge, let’s go to the[br]Thought Bubble.

0:07:53.760,0:07:59.589
During the Dartmouth Conference in 1956, the[br]most advanced computer was the IBM 7090.

0:07:59.589,0:08:04.789
It filled a whole room, stored data on basically[br]giant cassette tapes, and took instructions

0:08:04.789,0:08:06.860
using paper punch cards.

0:08:06.860,0:08:12.360
Every second, the IBM 7090 could do about[br]200,000 operations.

0:08:12.360,0:08:17.400
But if you tried to do that it would take[br]you 55 and a half hours!

0:08:17.409,0:08:21.479
Assuming you did one operation per second,[br]and took no breaks.

0:08:21.479,0:08:22.479
That’s right.

0:08:22.480,0:08:25.200
Not. Even. For. Snacks.

0:08:25.240,0:08:28.960
At the time, that was enough computing power[br]to help with the U.S. Air Force's Ballistic

0:08:28.960,0:08:30.120
Missile Warning System.

0:08:30.120,0:08:34.960
But AI needs to do a lot more computations with a lot more data.

0:08:34.960,0:08:39.720
The speed of a computer is linked to the number[br]of transistors it has to do operations.

0:08:39.720,0:08:44.600
Every two years or so since 1956, engineers have doubled the number of transistors that

0:08:44.600,0:08:46.760
can fit in the same amount of space.

0:08:46.760,0:08:49.190
So computers have gotten much faster.

0:08:49.200,0:08:54.720
When the first iPhone was released in 2007, it could do about 400 million operations per second.

0:08:54.760,0:08:58.560
But ten years later,[br]Apple says the iPhone X’s processor can

0:08:58.560,0:09:01.320
do about 600 billion operations per second.

0:09:01.320,0:09:05.480
That’s like having the computing power of[br]over a thousand original iPhones in your pocket.

0:09:05.490,0:09:10.370
(For all the nerds out there, listen you’re[br]right, it’s not quite that simple - we’re

0:09:10.370,0:09:11.990
just talking about FLOPS here)

0:09:11.990,0:09:17.550
And a modern supercomputer, which does computational functions like the IBM 7090 did, can do over

0:09:17.550,0:09:20.540
30 quadrillion operations per second.

0:09:20.540,0:09:25.710
To put it another way, a program that would[br]take a modern supercomputer one second to

0:09:25.710,0:09:31.790
compute,[br]would have taken the IBM 7090 4,753 years.

0:09:31.790,0:09:33.290
Thanks Thought Bubble!

0:09:33.290,0:09:37.200
So computers started to have enough computing[br]power to mimic certain brain functions with

0:09:37.200,0:09:43.760
artificial intelligence around 2005, and that’s when the AI winter started to show signs of thawing.

0:09:43.760,0:09:48.440
But it doesn’t really matter if you have[br]a powerful computer unless you also have

0:09:48.440,0:09:50.360
a lot of data for it to munch on.

0:09:50.360,0:09:54.390
The second development that kicked off the[br]AI revolution is something that you’re using

0:09:54.390,0:09:57.250
right now: the Internet and social media.

0:09:57.250,0:10:01.410
In the past 20 years, our world has become[br]much more interconnected.

0:10:01.410,0:10:05.510
Whether you livestream from your phone, or[br]just use a credit card, we’re all participating

0:10:05.510,0:10:06.650
in the modern world.

0:10:06.650,0:10:11.460
Every time we upload a photo, click a link,[br]tweet a hashtag, tweet without a hashtag,

0:10:11.460,0:10:17.280
like a YouTube video, tag a friend on Facebook,[br]argue on Reddit, post on TikTok [R.I.P.

0:10:17.280,0:10:22.660
Vine], support a Kickstarter campaign, buy[br]snacks on Amazon, call an Uber from a party,

0:10:22.660,0:10:25.720
and basically ANYTHING, that generates data.

0:10:25.720,0:10:29.510
Even when we do something that /seems/ like[br]it’s offline, like applying for a loan to

0:10:29.510,0:10:35.320
buy a new car or using a passport at the airport[br]those datasets end up in a bigger system.

0:10:35.320,0:10:40.260
The AI revolution is happening now, because[br]we have this wealth of data and the computing

0:10:40.260,0:10:42.130
power to make sense of it.

0:10:42.130,0:10:43.180
And I get it.

0:10:43.180,0:10:47.370
The idea that we’re generating a bunch of[br]data but don’t always know how, why, or

0:10:47.370,0:10:51.700
if it’s being used by computer programs[br]can be kind of overwhelming.

0:10:51.700,0:10:55.760
But through Crash Course AI, we want to learn[br]how artificial intelligence works because

0:10:55.760,0:10:58.320
it’s impacting our lives in huge ways.

0:10:58.320,0:11:00.650
And that impact will only continue to grow.

0:11:00.650,0:11:05.000
With knowledge, we can make small decisions[br]that will help guide the AI revolution, instead

0:11:05.000,0:11:08.320
of feeling like we’re riding a rollercoaster[br]we didn’t sign up for.

0:11:08.320,0:11:13.550
We’re creating the future of artificial[br]intelligence together, every single day.

0:11:13.550,0:11:15.360
Which I think is pretty cool.

0:11:15.360,0:11:20.440
Next time, we’ll start to dive into technical[br]ideas like supervised, unsupervised, and reinforcement

0:11:20.440,0:11:21.440
learning.

0:11:21.440,0:11:24.590
And we’ll discuss what makes a Machine Learning[br]algorithm good.

0:11:24.590,0:11:25.590
See you then!

0:11:25.590,0:11:28.360
Thanks to PBS for sponsoring Crash Course AI!

0:11:28.360,0:11:31.610
If you want to help keep all Crash Course[br]free for everybody, forever, you can join

0:11:31.610,0:11:33.880
our community on Patreon.

0:11:33.880,0:11:41.960
And if you want to learn more about how computers got so fast, check out our video on Moore’s Law.