Activa JavaScript para disfrutar de los vídeos de la Mediateca.
Estudiante del MIT: Historia de la IA. - Contenido educativo
Ajuste de pantallaEl ajuste de pantalla se aprecia al ver el vídeo en pantalla completa. Elige la presentación que más te guste:
Presentación sobre la historia de la IA y su futuro a cargo de estudiante del Instituto Tecnológico de Massachusetts (MIT, Massachusetts Institute of Technology). Programa MIT Global Teaching Labs. IES Gran Capitán.
Okay, we have about 15 minutes, so I'm going to go through this more quickly, but I just
00:00:00
want to talk a little bit more about the history of AI, because that's the hot topic these
00:00:10
days, and I think we should all learn a little bit more about what it is instead of using
00:00:15
it as a buzzword.
00:00:19
Cool.
00:00:21
All right, so what is AI, right?
00:00:23
All right, AI is just a machine, like the intelligence of machines or programs.
00:00:24
So that sounds simple, but let's unpack that, right?
00:00:31
So there's multiple types of AI.
00:00:36
There's narrow AI, and there's general AI.
00:00:37
So narrow AI is you specialize in a specific topic, and general AI is you can learn to
00:00:40
do and solve any problems, just like a human.
00:00:50
So here are some examples of AI, self-driving, go-playing or chess-playing board games, and
00:00:54
finally, robots that can walk and run, just like humans.
00:01:04
Okay, great.
00:01:10
So I'm going to skip some of this, because this is some technicalities, but basically,
00:01:11
the precursors of AI is how do we formalize human thoughts?
00:01:16
How do we formalize?
00:01:22
Human thought in a mathematical way, and so there are many theorems that came up in
00:01:24
the 1930s that formalized the math of human thought, and in 1945, we got the first computer
00:01:29
that was programmable and electronic and digital, so it was this big.
00:01:41
So think like this whole classroom, that's how big it was, and you had many people who
00:01:46
had to operate this machine in 1945.
00:01:51
Great.
00:01:54
So now let's talk about some of the fundamentals of AI.
00:01:55
So people started talking about how we can create an artificial brain, and so there's
00:02:01
a Turing test.
00:02:08
Maybe you've heard of this, but this is just something, a test that can tell if a computer
00:02:09
can say intelligent things like a human.
00:02:15
And so we also got some really cool programs.
00:02:23
Like the first neural network machine built by Marvin Minsky, a MIT professor for a long
00:02:24
time, and some other programs of AI, for example, logic theorists that proved many math theorems,
00:02:33
which is kind of insane to think about back in the 50s.
00:02:40
Great.
00:02:46
So this is where it all started.
00:02:47
The birth of AI.
00:02:48
1956, the Dartmouth workshop.
00:02:49
Dartmouth is a university in the United States.
00:02:51
This is where AI got its name, got its first success, and these ten folks were some of
00:02:53
the people who attended, and they became famous for their contributions to AI.
00:03:01
We saw earlier Marvin Minsky, and Cloud Shannon is also very famous for his contributions.
00:03:06
And everyone is, but those are the two that I personally know the most because they're
00:03:14
from MIT.
00:03:18
Great.
00:03:19
Thank you.
00:03:20
Thank you.
00:03:21
Great.
00:03:22
So in this time range, we had many advances.
00:03:23
So computers were able to solve algebra problems, geometry problems, learn English, just some
00:03:28
crazy things that were going on.
00:03:35
And so a lot of these universities like MIT, CMU Stanford, they received millions of dollars
00:03:37
from the government to research.
00:03:43
And so we were thinking, or not we, they were thinking that in 20 years, we would have a
00:03:46
computer as smart as a human.
00:03:51
And so, yeah.
00:03:53
Like, DARPA was just feeding money to these universities.
00:03:54
And so maybe not to get too complicated here, but back in the 50s, the AI was more about
00:04:01
searching.
00:04:09
So you have some big space of possible choices or possible solutions, and you have to search
00:04:10
and you have to search it to find the best one.
00:04:19
And we also got the first kinds of neural networks,
00:04:26
if you've heard of that.
00:04:30
Those are the most popular type of AI today.
00:04:30
And so we had very small neural networks,
00:04:34
very simple ones, back in the 50s.
00:04:38
This is where it all started.
00:04:40
And we also got the first chatbot created by,
00:04:43
yes, another guy from MIT.
00:04:46
It was created to be some sort of psychotherapist,
00:04:47
in some sense.
00:04:54
So let's look a little bit more at the text here.
00:04:55
I'll leave this later, but just read it briefly.
00:05:00
And it's quite impressive to me,
00:05:05
this chatbot that was made 50 years ago.
00:05:08
It's giving some very human-like responses.
00:05:12
Okay.
00:05:16
So, all right.
00:05:17
Of course, we did not make the general AI in 20 years.
00:05:21
Instead, we got the first AI winter.
00:05:25
So there were some fundamental barriers
00:05:28
to actually implement AI.
00:05:32
So, for example, there's something called
00:05:35
combinatorial explosion.
00:05:37
That's just a bunch of fancy words that really means
00:05:38
when your problem becomes complicated,
00:05:41
the solution space becomes,
00:05:45
really, really big.
00:05:47
And so another problem was the computers in the 50s,
00:05:54
70s, excuse me, were terrible.
00:05:57
Like, look at the Cray 2 supercomputer,
00:05:59
super computer, back in 1985.
00:06:02
And look at an Apple iPhone 6 smartphone.
00:06:05
The comparison is astounding.
00:06:08
Our Apple iPhone 6 is hundreds of times better
00:06:10
than this super computer back in 1985.
00:06:13
And finally, most programs in the 70s
00:06:17
didn't have common sense.
00:06:22
So they didn't understand the things that a baby human would,
00:06:24
like, how do you throw a ball?
00:06:30
How do you look at digits?
00:06:33
They just did not have common sense.
00:06:34
So, of course, everyone lost their funding.
00:06:39
The government said, you guys are not doing well.
00:06:41
And so, unfortunately, also, you know,
00:06:44
the networks, the research there and funding there
00:06:47
also got cut off because Marvin Minsky,
00:06:50
same guy from earlier, and another person named Poppert,
00:06:54
both MIT folks, they published a book that said,
00:06:57
hey, neural networks don't work.
00:07:00
And so they basically just, yeah, it was bad.
00:07:02
And so now after the first decline, the first winter,
00:07:06
we had actually an AI boom.
00:07:13
So thankfully, things recovered in the 80s.
00:07:15
And so there's these things called expert systems.
00:07:17
So people started working on expert systems
00:07:23
where instead of trying to solve all of AI,
00:07:26
they started working on narrow AI,
00:07:29
just solving a very small portion of AI problems.
00:07:31
And so an expert system would become an expert
00:07:34
on a small topic.
00:07:36
And then other people, not experts,
00:07:38
could ask the expert system, how do I do this?
00:07:40
Can you help me do this, right?
00:07:44
Kind of like ChatGPT,
00:07:45
but just for a very small dominion.
00:07:47
So like just about pasta or just about making pizza
00:07:50
instead of the whole world.
00:07:53
So then knowledge became the primary focus of research.
00:08:00
How can we gather information
00:08:06
and how can a robot synthesize information and understand it?
00:08:09
And of course, funding came back.
00:08:15
So Japan started first, and then the United Kingdom
00:08:17
and the US followed with millions of dollars of funding.
00:08:20
Also, neural networks, they came back
00:08:25
because a very smart physicist demonstrated
00:08:28
that neural networks can actually work.
00:08:31
They can converge.
00:08:34
And they can, another very important computer scientist,
00:08:35
Hinton and Rumelhart, they popularized something
00:08:39
called back propagation that is today still one of the most popular ways to do this.
00:08:43
The most popular way for the world to train AI.
00:08:47
And of course that didn't last too long.
00:08:51
Another second winter came, unfortunately.
00:08:53
What happened was there was a collapse in the market
00:08:56
for specialized AI hardware,
00:09:01
because Apple and IBM, they started selling computers
00:09:03
that beat the AI machines.
00:09:07
So if you remember in 1987,
00:09:11
that was when the Apple II personal computer was released.
00:09:13
And that was a huge big deal.
00:09:16
revolution and the personal computing space and also AI expert systems they work sometimes but
00:09:17
usually they stop working after you push them to their limits and so this is the Apple computer
00:09:26
Apple too and so the United States government said we don't like AI anymore we're going to
00:09:32
take your money and put it elsewhere and so also a lot of AI companies they lost they lost their
00:09:41
funding they closed or they were sold to other companies now something bright that came out of
00:09:49
this was people realized maybe there's actually we're doing this the wrong way maybe instead of
00:09:56
thinking about how smart a human is we should think about how good a human as as understanding
00:10:02
the world so instead of building brains we should
00:10:08
you know we should think about how smart a human is we should think about how good a human is
00:10:10
build bodies so some smart researchers thought that okay instead of building the brain of the
00:10:11
machine let's just build a machine that can learn how to perceive the world instead of just looking
00:10:20
at words but instead of thinking about how do I hold a tennis racket how do I look at the room
00:10:26
and understand that I'm in a room so basically advanced robotics cool
00:10:33
all right now narrow AI started having some success in the 90s and the early 2000s for
00:10:41
example Deep Blue it beat the chess champion Garry Kasparov in 1997 a very famous example
00:10:47
of machine beating human also some successes in autonomous driving some robots from Stanford and
00:10:55
CMU navigated a car with no driver in the early 2000s for hundreds of millions of dollars and
00:11:04
kilometers and also if you know the quiz show trivia show jeopardy IBM Watson was able to beat
00:11:11
the two best human champions of all time by $50,000 in 2011 of course these successes were not because
00:11:21
humans made better algorithms it was actually because computers just got way better so there's
00:11:33
something called Moore's law where there was a
00:11:38
tremendous uh pattern of increase in computer capacity in these uh in these years this is an
00:11:40
example of Moore's law you can take a look at it later but it just shows you that progress
00:11:52
was tremendous it was exponentially good and so the AI had at this point AI had resolved many
00:11:57
problems but it was still not considered uh a fun area because AI had two winters and so people
00:12:10
didn't want to say that they were working at AI but its applications were all across the industries
00:12:22
so basically AI very useful but nobody wants to call the work AI because it was not the cool topic
00:12:32
it was the weird topic
00:12:39
very different from now right okay now very recently you might remember the big data
00:12:40
bubble or the hype basically big data is you have a lot of data and you need very big computers to
00:12:49
work with a lot of data and we can do that because we have cheap computers lots of computers and lots
00:12:55
of data so also we had some Advances in deep learning
00:13:01
where we got some really uh crazy deep uh deep learning algorithms and also had the compute
00:13:10
capacity to actually use them to solve problems like identify cats and dogs or identify what kind
00:13:20
of object where the object is in an image great so let's talk about the present 2017
00:13:27
our transformer
00:13:39
um that was so we had this thing called GPT-3 that's the precursor of the um that's a big
00:13:40
it's a large language model that was based on the transformer architecture and GPT-3 you see GPT
00:13:52
well yeah chat GPT was based on GPT 3.5 um and that was launched in 2022
00:13:59
and also uh now
00:14:09
Microsoft is working on GPT-4 which is supposed to be even better and there are some chat GPT
00:14:10
versions based on GPT-4 but you have to pay money to see them cool so um this is a small
00:14:16
demonstration of some current research in AI some really smart Stanford researchers they made a
00:14:25
robot that can cook food for you um to be fair this one is controlled by the human but their robot also
00:14:30
cooks stuff on its own too autonomously um but yeah it's just cooking some uh Cantonese food
00:14:39
um I don't know if it's any good or not but I'd be down to try it but we can skip it um
00:14:46
all right cool all right okay let's talk about AI ethics right this is important because when you do
00:14:53
you need to make sure that you are not harming people all right how do we
00:14:59
eliminate bias in science examples Amazon tried to once use a AI system to
00:15:04
recruit people and try to interview people and it's not using it why because
00:15:11
the algorithm only chose men because mostly men work at Amazon so it assumed
00:15:18
that all the best engineers are men which is obviously not true right
00:15:23
another problem facial detection it had more data from white men and less data
00:15:26
from darker woman so therefore it was worse on detecting darker woman so it's
00:15:33
you know this is not okay right we need to have a unbiased system for
00:15:38
classifying or detecting things another problem is job loss AI can do a lot of
00:15:44
things that humans can do
00:15:55
you
00:15:56
better faster and people might lose jobs and so we need to make sure that people
00:15:56
are able to still be able to make enough money to live if their job gets taken
00:16:02
away or have an alternative job that they can do instead and finally this is
00:16:09
a little bit far out there I don't think it's going to happen but some people do
00:16:15
existential risk what if we make an AI that's so smart
00:16:17
you can't stop it and it like kills all humans right that I don't ask what
00:16:22
happened but hey I didn't say that all right so finally I like to show a
00:16:26
timeline you can click on this link when I share the position afterwards and look
00:16:34
at some of the advances that we've talked about and yeah so that was a
00:16:38
little bit of an intro thank you again for listening and once again my email
00:16:44
and any questions I'd be happy to take thank you
00:16:47
you
00:16:51
you
00:16:53
- Idioma/s:
- Idioma/s subtítulos:
-
- Autor/es:
- Rafael M
- Subido por:
- Rafael M.
- Licencia:
- Reconocimiento - No comercial
- Visualizaciones:
- 32
- Fecha:
- 3 de febrero de 2024 - 10:27
- Visibilidad:
- Público
- Enlace Relacionado:
- https://docs.google.com/presentation/d/1Nv7phRAF5OQ5snWBEGHACJ8AW5SEEyUgDeFYYD-rcEU/edit?usp=drive_link
- Centro:
- IES GRAN CAPITAN
- Descripción ampliada:
- Historia IA
- Duración:
- 16′ 59″
- Relación de aspecto:
- 1.78:1
- Resolución:
- 1280x720 píxeles
- Tamaño:
- 492.61 MBytes
