Placeholder Image

字幕列表 影片播放

  • So, we have six pieces of art right here.

  • Five of them have been created by a human

  • and one of them has been created by a machine.

  • Which one is the machine-made piece of art?

  • We're gonna take a vote.

  • The first one, the second one, the third one, the fourth,

  • the fifth and the sixth.

  • No one got it right.

  • NARRATOR: The School of Life is sending some of YouTube's most popular creators

  • on a series of field trips

  • to explore some of philosophy's most intriguing ideas.

  • This time, it's Luzu,

  • a Spanish content creator who built a global reputation

  • playing online video games

  • and now has over eight million subscribers.

  • Advances in artificial intelligence

  • are blurring the lines between humans and machines.

  • And now robots are encroaching on the last remaining area

  • that we thought was uniquely human.

  • The power to be creative.

  • I'm Luzu. Or am I?

  • The one thing that I keep thinking about is,

  • as we give these machines human-like attributes,

  • what does that mean

  • when it comes to our identity as humans?

  • This is such a big subject,

  • we need a little guide before we start.

  • NARRATOR: A 60-second guide to identity.

  • For most of human history,

  • the difference between humans and machines was very clear.

  • A machine was something clever but simple,

  • whereas humans were miraculous creatures of astonishing potential.

  • But very recently, something extraordinary has happened.

  • We've started to make machines that equal and outstrip many of our own capacities.

  • This has created a troubling question.

  • What is left that makes us distinctively human?

  • One important answer is the ability to create.

  • Machines might be better at predicting the weather,

  • or even beat us at chess,

  • but surely they'll never be able to generate amazing art.

  • In 1950, Alan Turing, the founding father of computer science,

  • proposed a test.

  • When a human can no longer tell

  • if they are interacting with a person or a machine,

  • we will have lost the one last quality

  • that distinguishes us from machines.

  • So, when we can no longer tell

  • the difference between art made by machines

  • and art made by humans,

  • what would be left to mark us out as unique?

  • Might as soon be nothing at all

  • that we can do that a machine won't be able to do just as well?

  • Or, most frighteningly of all, much, much better.

  • Machine learning is now challenging

  • our very concept of what makes humans unique.

  • We're comfortable with robots performing basic skills

  • that we don't want to do ourselves,

  • but in this video,

  • Luzu was looking at AIs that create art

  • before facing a test, along with you,

  • to see if he can correctly identify art

  • created by a machine

  • when it's up against art created by humans.

  • But first, what does life look like on the AI frontline?

  • Hi, there. I'm Luzu. Nice to meet you.

  • Nice to meet you, too. How are you?

  • I'm great. Thanks for asking.

  • NARRATOR: This robot is called NAO.

  • It looks a bit like a toy,

  • but, actually, it's a WiFi enabled, autonomous AI

  • who listens through sensors and gathers information from online sources,

  • teaching itself through a continuous process of machine learning.

  • NAO gives an eerie sense of how the border

  • between human and machine is becoming more porous.

  • -I need to learn some lines. -[EXCLAIMS]

  • -Don't distract me. -Oh.

  • I'm here trying to figure out what it means to be human.

  • Do you know anything about that?

  • Absolutely. We robots are evolving quickly

  • and taking on more and more of your human characteristics.

  • We have been able to carry out straightforward tasks,

  • but recently we have begun to educate ourselves.

  • It's called machine learning.

  • Think I'm gonna want to research

  • and learn a little bit more about that.

  • NARRATOR: We humans think of ourselves as the undisputed masters of the universe,

  • but that confidence is now being challenged

  • as advances in machine learning

  • close the gap between what humans and computers can do.

  • There is one area where we feel certain

  • that we do still retain a unique distinction.

  • Only humans have the power to be creative.

  • But what if even that skill

  • could be performed by a machine?

  • Before being tested on his ability to spot the art created by an AI,

  • Luzu's on his way to UCLA,

  • where an AI is teaching itself

  • to create an original artwork.

  • So, this is one of the projects I wanted to show you.

  • The concept is really to teach a machine

  • to create a single portrait

  • from this vast number of individual portraits.

  • And then asking it to generate its own portrait from that.

  • LUZU: The machine creates something original, right?

  • It's not just imitating everything,

  • it's creating something from scratch.

  • GABBY: So, there's two parts of the algorithm,

  • one of them generates an image and the other one

  • criticizing it and gives a little bit

  • of feedback and says, "Actually,

  • "you need to make it a little bit smaller,

  • "or put a bit of red in this area and try again."

  • So, as the process goes on, it's getting better and better.

  • So, this shows the process of evolution,

  • where on the left we got some of

  • the early imagery that's generated by the machine

  • and it starts to become more and more

  • kind of recognizable as one of the portraits from the set.

  • But it still has a very painterly, slightly abstract,

  • slightly Impressionistic feel.

  • But this was just running for 24 hours.

  • Makes me think a little bit of

  • how a baby learning what a person is.

  • The machine is right now doing the same thing.

  • The machine learning processes are so similar

  • to our own learning processes and our creative processes.

  • That you start to see that kind of invention

  • and creativity happening.

  • I think in the same way that has been happening within humans.

  • So, the machine's kind of going through that process

  • at an accelerated rate.

  • NARRATOR: It's one thing for machines to create original art,

  • but is it any good?

  • AI company GumGum is putting art made by machines

  • up against art made by humans

  • to see if Luzu can tell which is which.

  • We're deeply interested in computer vision

  • and machine learning and we decided, why don't we try

  • a Turing test?

  • And see if humans can figure out

  • which one's artificially generated.

  • I know that art is very subjective

  • and something that we would attribute to, like, humans.

  • There's a lot of emotion, there's expression,

  • so how does a machine choose what it's gonna paint?

  • I mean, I can flip that back on you

  • and say, "How do you decide?"

  • -Is it based on data? -I guess...

  • Computers are more than capable of generating art

  • in any medium.

  • From all these paintings,

  • one was created by a machine?

  • That's right, yeah...

  • Any idea which one it might be?

  • So, it could be any of them, right?

  • Right. Which one is it?

  • NARRATOR: While Luzu ponders,

  • why don't you have a go?

  • Can you tell which one of these paintings

  • has been done by an AI?

  • Do you think it's one?

  • Two, three?

  • Four?

  • Five?

  • Or six?

  • If I had to take a guess,

  • I don't know a lot about art or anything like that,

  • but I think I would choose the first one.

  • Just because,

  • it looks a little bit more flat, right?

  • I don't know, it looks like something that

  • could be made by a machine.

  • It's like the randomness that a machine would...

  • Yeah, I imagine something like that.

  • -Like zeros and ones. -Yeah, kind of. [CHUCKLES]

  • Actually, it's this one.

  • LUZU: Okay, I was not expecting that.

  • The way that it works is actually by

  • taking images like this

  • and teaching it how to generate things of this nature.

  • And, eventually, after you feed it enough data,

  • you'll yield something that looks like this,

  • that passes a Turing test,

  • which actually blew my mind. I was so happy.

  • The definition of creativity is to create.

  • Doesn't say anything about being a human or a monkey or machine.

  • So, try and answer the question where this will end,

  • these things are just gonna keep getting bigger,

  • faster, stronger, better, and whether the data comes from us

  • or maybe they start learning from each other,

  • I think it's a limitless ceiling.

  • LUZU: Do you see that as a possible thing, we end up being the ones

  • -learning from? -Absolutely, and the fact that

  • you couldn't discern is a testament to that.

  • It validates the question.

  • NARRATOR: Until recently, machines from the oldest cameras

  • to the most modern digital scanners

  • have only been able to capture impressions of reality

  • created by humans.

  • Now, machines are doing something far more interesting.

  • They're developing distinctive artistic personalities

  • in the same way as Van Gogh, Picasso or Salvador Dali.

  • Machines can do simple, repetitive tasks like frying burgers,

  • and complex ones like beating us at chess.

  • But if it will soon be making art

  • good enough to hang in museums,

  • we may have to rethink what claims we can still make

  • to have any unique or distinctive qualities.

  • And decide whether to feel alarmed by

  • this threat to our concept of humanity,

  • or proud of having built machines that will soon be our equals.

  • Or perhaps even our betters.

  • Although they may not be quite ready

  • to replace us just yet.

  • [YELPS]

  • You were trying to go too far.

  • Like, I know I'm not that flexible, either.

  • [YELPS]

  • All right, I got you. Come on, my man.

  • -Come up, you got it? -[WHIRRING]

  • I think you got it.

  • Okay? You're almost there.

  • That was amazing.

  • Can you high five?

  • Not there yet? Okay.

  • Thank you, guys, so much for watching.

  • If you enjoyed it, what can they do?

  • Make sure you subscribe to the School of Life channel

  • to see the rest of the series.

  • Thank you, and thank you, guys, for watching.

  • [INDISTINCT CONVERSATION]

So, we have six pieces of art right here.

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

A2 初級

下一個畢加索會是機器人嗎?| 呂祖學派 (Will The Next Picasso Be A Robot? | The School of Luzu)

  • 2 0
    林宜悉 發佈於 2021 年 01 月 14 日
影片單字