LOADING ...

These Google AI experiments are crazy! This is the future.

1M+ views   |   43K+ likes   |   839 dislikes   |  
Mar 08, 2017

Thumbs

These Google AI experiments are crazy!   This is the future.
These Google AI experiments are crazy!   This is the future. thumb These Google AI experiments are crazy!   This is the future. thumb These Google AI experiments are crazy!   This is the future. thumb

Transcription

  • Hi! I'm Dave from Boyinaband.
  • Today we're gonna take a look at a bunch of artificial intelligence experiments
  • To see how far along we are towards our imminent doom.
  • First thing I wanna take a look at is right here.
  • Google has this aiexperiments.withgoogle.com site
  • Where they have tons of these things, and...
  • Well, nine.
  • [laughing] Uhm.
  • And there's some pretty fun ones
  • So, the first one I want to show you is this one
  • "Quick, Draw!"
  • So, basically, it's like Pictionary, except it's an artificial intelligence
  • So Pictionary if you have no friends
  • Can a neural network learn to recognise doodles?
  • Help teach it by adding your drawings to the world's largest doodle data set
  • Which could be shared publicly to help with machine learning research in the future.
  • So, by playing this we're actually improving the A.I.!
  • [Ominous music]
  • So the whole point is we gotta draw things as quickly as possible
  • And it has to guess it, using machine intelligence
  • Which is really cool
  • Yeah, the only issue here is my drawing ability hasn't progressed since age five
  • Let's try it
  • Uhhh, helicopter...
  • [Machine: I see circle]
  • [Machine: Or pear]
  • A pear? [ha ha]
  • [Machine: Or pineapple]
  • [Machine: I see drill]
  • [Machine: Oh, I know, it's helicopter!]
  • A drill?!
  • It got that!
  • [Helicopter sounds]
  • [Helicopter whirring]
  • [Helicopter choppa sounds]
  • How the hell?
  • Draw a bear
  • In under twenty seconds, okay...
  • They have those little noses
  • [Machine: I see circle]
  • [Machine: Or lollipop]
  • I see circle...
  • [Machine: I see camoflague]
  • [Machine: Or peas]
  • [Machine: Or dog]
  • [Machine: Oh, I know, it's bear!]
  • It got it!
  • How does it know?
  • It's so cool
  • Broccoli...
  • [Whispering] I really like broccoli...
  • How will it know this apart from tree?
  • [Machine: I see matches]
  • Oh my god...
  • [Machine: Oh, I know, it's broccoli!]
  • WHAT the HELL?!
  • How did it assume broccoli before tree?
  • We're all dead
  • How the hell
  • Like
  • A human wouldn't be able to do that
  • Draw a broom, alright...
  • [Machine: Oh, I know, it's broom!]
  • [Surprised laughing]
  • [Inhale]
  • Oh my god
  • [Machine: I see line]
  • [Machine: Oh, I know, it's streetlight!]
  • WHAT on Earth...
  • [Laughing]
  • I'm...
  • I'm really shocked
  • That it managed to get that
  • I want to try that again
  • That's crazy
  • Okay...
  • A bee...
  • [Machine: Oh, I know, It's bee!]
  • I don't eve-
  • Asparagus??
  • [Laughing] Shocking that it didn't recognize that
  • Uhm, so yeah, that's what asparagus looks like
  • You can tell how much veg I eat
  • Oh my god, someone drew that as a cooler?
  • It looks like a dead cat
  • A dead, square cat
  • A blackberry
  • Oh, that's cute
  • Asparagus...
  • I'm not certain I would call that asparagus
  • What other bees did people draw?
  • Aw, that one's adorable
  • Yoga??
  • Oh, that's a hilarious one
  • This is like a way of making visual puns
  • How about a cup?
  • Yep, everyone went for the same thing
  • Ooh, nice cup of tea
  • That's made me wanna get a cup of tea
  • [Clinking noises]
  • [Click]
  • Think I'm gonna trigger some people right now
  • Milk before the water
  • *How much sugar Dave? Are you okay?*
  • Oh hell yeah
  • [Slide]
  • [Clunk]
  • Recycle
  • [Laughing] I wouldn't be disappointed if this became a staple part of my videos
  • [Clap]
  • Where were we?
  • Oh yeah
  • Imminent doom
  • Let's move on
  • What other experiments d'y-you have for us?
  • Giorgio cam
  • Use your camera to make music with me!
  • Okay
  • Alright...
  • [Machine: Ready to rock image recognition with me?]
  • [Machine: Take a picture and I'll tell you what I see]
  • [Drum beat and Electronic music]
  • [Machine: Snap a photo]
  • [Drum beat and Electronic music]
  • [Click]
  • [Machine: H-H-Here we go]
  • What the hell is this?
  • [Machine: Alright]
  • [Machine: Working on it]
  • [Machine: Almost there]
  • [Beat drop]
  • [Electronic music and Drums]
  • [Machine: Is that Glasses or not?]
  • Yes it is!
  • [Machine: Got Face in this spot?]
  • [Laughing]
  • I dunno!
  • This is awesome!
  • What the hell is this??
  • [Electronic music]
  • [Machine: Almost there]
  • [BREAKDOWN]
  • Awwh that breakdown
  • Yes it is hair
  • [Machine: Got Beauty in this spot]
  • [Laughing] Thank you!
  • [Electronic music]
  • This is crazy! Whoa!
  • [Electronic music]
  • [Machine: Take another picutre]
  • This is so worth Google's time!
  • [Click]
  • [MLG Airhorn lol]
  • [Laughing] Airhorn!
  • [Machine: Ya ya ya ya]
  • [Breakdown]
  • [Machine: Working on it]
  • [Machine: Looks like Single Lens Reflex Camera to me]
  • What the hell?!
  • [Machine: I'm seeing Camera probably]
  • [Electronic music]
  • [Machine: One more time]
  • What the hell?
  • The specific type of camera?
  • What will it do with the background?
  • [Machine: Alright]
  • [Machine: Here we go go go go go]
  • Airhorn!
  • [Machine: That's Library on my screen]
  • Yes it is!
  • [Machine: I think furniture is what I've seen]
  • [Laughing]
  • [Machine: That was awesome]
  • [Cheering and applause]
  • [Machine: I am shutting down]
  • [Cheering and applause]
  • I...
  • [Hah]
  • I am actually speechless
  • Oh my god
  • I am so glad I didn't try that
  • Prior to recording
  • That's amazing
  • Eric Rosenbaum and Yotam Mann
  • Follow
  • Follow
  • You guys are fricken' brilliant
  • Looks a bit like Andy Samburg
  • How does it work?
  • Man: "There are two parts
  • Image recognition for seeing things
  • And speech synthesis, for talking
  • For image recognition, we used the Cloud Vision API
  • You can just send it a picture
  • And it returns labels for what it sees
  • And for the computer's voice
  • We used an open source project calls Mary TTS
  • Combined with the Web Audio API
  • This let us add effects to the computer's voice
  • And sync it to the beat."
  • That's so cool!
  • I'm so freaking speechless at how cool that is!
  • Oh, man...
  • That got me so hyped!
  • This artificial intelligence can rap [laughing]
  • Amazing!
  • I'm out of a job
  • It can- not only can it rap
  • It can freestyle!
  • It can freestyle better than I can!
  • Let's head back to the experiments
  • Oh, that's so good!
  • Ok, the Infinite Drum Machine, what've we got here?
  • Create beats using sounds from the every day world
  • So basically this is an Andrew Huang simulator?
  • [Slap click slap click] Drop that beet!
  • Alright, start playing
  • [CLANG]
  • Ah!
  • [Clang, click]
  • Oh, we can move it around
  • Awh, that's so cool!
  • [Multiple clangs, clicks, sounds etc.]
  • Alright, what's the beat sound like as it is, we've got...
  • [Fzz] Paper bag small open and remove
  • Billiards cue [clump]
  • [Pf] Clothes drop]
  • [Multiple sounds played in order]
  • Now, let's uh, let's try and get a- a decent kind of four on the floor to start
  • [Dave beatboxing]
  • Kinda thing
  • [Multiple sounds]
  • This is so cool!
  • This is such a fun way to make music!
  • [A lot of noises playing at once]
  • It's kind of ordered by frequency
  • So like, over here is all high pitch stuff [multiple noises in the background]
  • [Lots of noises]
  • Then you get the middle [more sounds]
  • And then the deep [more noise]
  • The deepest [bvv] Awh!
  • [Playing the deepest sounds]
  • Awh, that's gorgeous!
  • They've sorted the samples by like, frequency content. That would be so useful in a music programme!
  • Animal Lion... Oh, and they've got different tags and stuff!
  • [Rapping]
  • Ah, we can change the tempo!
  • [Fast beat] Raise yo fists!
  • Can we get a [beatboxing an example]
  • [Noise]
  • Yeah!
  • Ah, and it's got all these different thi- noise makers? What classes as a noise maker?
  • [Drums and trumpet] - [Dave laughs]
  • [Noise]
  • Metal drop?
  • [METAL GUITARS]
  • This is fun.
  • Alright, Thing Translator - Take a picture of something to hear how to say it in a different language
  • [Machine: Cabello]
  • Hair...
  • I am just hair.
  • [Laughs]
  • Judging from the views on my recent video about it, YouTube things so too.
  • Hair 98%, face 97% [laughs]
  • It's more sure that I am hair than I am a face.
  • Alright, uh, what's this? What's this?
  • [Machine: Beber. Drink.]
  • [Posh accent] Beber.
  • Justin drink is my FAVOURITE pop singer.
  • Ah, brilliant, let's learn us some Japanese!
  • [Funny voice] I'm gonna be a Japanese one day!
  • It's the key to success. C'mon, Japanese people.
  • [Machine: Hakimono. Footwear.]
  • No... y'... I dunno where you're getting that.
  • [Machine: *something Japanese*. Nail.]
  • [Machine: *Japanese*. Art.]
  • No, you're not... it's a key. IT'S A KEY.
  • [Machine: *Japanese*. Finger.]
  • UHHHHHHHHHH
  • If only it was rapping. The rapping one was way more effective.
  • I swear it must use the same API though.
  • Can you get a hairbrush from the back?
  • Product [laughing]
  • VERY SPECIFIC *sarcasm intensifies*
  • How 'bout ma cuppa?
  • [Machine: *Japanese*. Glasses.]
  • Okay, yeah, okay...
  • C'mon, just guess it's a cup!
  • Oh my God!
  • Look. I'll- I'll make this eas- I'll make this SUPER easy.
  • [Machine: GADGETO! Gadget.]
  • Gadget?!
  • This feels like doing a video with my sister
  • [Machine: Haiya. Hair.]
  • Well, at least we know what 'hair' is in Japanese now.
  • Alright, I'm not feeling this one as much as some of the others.
  • Thousands of bird sounds visualised using the machine learning
  • [A lot of birds]
  • Hooo
  • [More birds]
  • This is weird.
  • Ooh, here we go.
  • Visualising high-dimensional space?
  • This sounds like one of those fricken' unscientific spiritual YouTube videos that are like:
  • [Silly voice] "How to visualise your highest dimension!"
  • That stuff bothers me.
  • So this is like a 3D visualiser that automatically maps things together depending on different factors
  • Fetching sprite image, computing PCA...
  • Have so much idea what these things mean.
  • Oh, what the hell?
  • [Clunk]
  • It's pretty.
  • Why... what are all these numbers?
  • Why... why.
  • Uhm. I'm actually really interested in this, how does it work?
  • Oh my God, I am so lost.
  • Yeah, I'm not sure that I have time to really start to understand this [laughing]
  • Just yet.
  • It would be fun to really sit down and try and get this
  • Let's just watch the video, see if that explains.
  • Person: 'Hi, I'm Daniel.'
  • Other person: 'Hi, I'm Martin.'
  • Person #3: 'Hi, I'm Fernanda.'
  • Hiiiii!
  • Person: 'We're looking at a subset of words that the computer has learned about.'
  • Person: 'We can also look closely at smaller sets of words...
  • ... if we search 'piano', we can run [??] the only words related to piano.
  • We get clusters of composers, genres...'
  • That's so cool!
  • Person: 'Each image is 784 pixels. The computer treats each pixel as a dimension.'
  • That's what all the numbers were.
  • Okay, so this is all the words.
  • So what happens if I search for like...
  • [Typing]
  • Dre [laughing]
  • They don't forget about Dre.
  • Wow - rap is more associated more with Dre than it is with musician.
  • That's pretty impressive.
  • I feel like I understand this well enough to do a very interesting part on it.
  • I think we're gonna... end it as- as it should be. On a sick rap.
  • [Machine: Ready to rock image recognition with me?]
  • [Machine: Take a picture and I tell you what I see]
  • [sigh] This is it [click]
  • [Machine: H-H-Here we go]
  • [Machine: Is that Clothing I see?]
  • Yes! It is! Alright, we got another clothing? [?]
  • Let's- let's step it up a little
  • Gonna get you, Cry
  • [Machine: Working on it.]
  • [Machine: Almost there.]
  • [Machine: IDK]
  • [Hysterical laughter]
  • [Machine: Take another pic okay?]
  • [Herp]
  • [More laughing]
  • Alright, so it has its limits.
  • That is fricken' crazy. I can't believe how cool machine learning is.
  • Awh, it really makes me wanna learn more about how it works.
  • I- I get fricken'- I nerd out on this stuff a lot [laughs]
  • Cheers for wat- for joining me and this- let me know if you want me to do any more videos on AI things.
  • Oh, if you know any tools like this kinda thing, let me know 'cause I bet there's tonnes of things like this that no one really knows about
  • on like, obscure sites, like that edges-to-cats thing and stuff like that.
  • Man... that was so much fun. It sucks that it's probably gonna kill us all.

Download subtitle

Description

Google has some insane experimental AI - I figured it was worth exploring to see how close it is to destroying us all.
Check them out yourself: https://aiexperiments.withgoogle.com

► ► If you want to support what I do, this is the best way: http://www.patreon.com/boyinaband
► Facebook: http://www.facebook.com/boyinabandcom
► Twitter: http://twitter.com/DavePBrown
► Tumblr: http://davebiab.tumblr.com/
► Music: http://boyinaband.bandcamp.com
► T-Shirts: http://boyinaband.bigcartel.com/
★ ★ ★

Trending videos