Latest / kill switch / Chocolate Chicken Chicken Cake
Transcript
- 0:04Speaker 1: Sleepwalkers is a production of our Heart Radio and unusual productions.
- 0:12Speaker 1: So I'm here for a surprise poetry reading. It's about
- 0:16Speaker 1: to start. The silence is hardly final. Somewhere in the street,
- 0:21Speaker 1: I can see the trees begin to rise and fall
- 0:24Speaker 1: for the light of the dark thing above me. The
- 0:27Speaker 1: dream is like a shiny black hair, and the sun
- 0:29Speaker 1: is like a dream. I stand up and watch the
- 0:32Speaker 1: sun shine on a single day, and the sun has
- 0:35Speaker 1: a chance to accomplish from the springs of my own delight.
- 0:40Speaker 1: Kind of haunting, abstract, yes, but beautiful too. And crucially,
- 0:47Speaker 1: when I read this, I felt, as you just did.
- 0:50Speaker 1: I hope that it is beautiful. I found it evocative
- 0:53Speaker 1: of experience that I've had it in the past. I
- 0:55Speaker 1: found it nostalgic um and I'm like, oh my god,
- 0:58Speaker 1: I'm having real human being emotions. That was filmmaker Oscar Sharp,
- 1:02Speaker 1: and that poem wasn't written by him or by anyone else.
- 1:06Speaker 1: It was written by a computer, a machine poet. We're
- 1:10Speaker 1: more and more worried about robots coming to take our jobs.
- 1:13Speaker 1: And though perhaps few would regret less hips to poets
- 1:16Speaker 1: shambly around Brooklyn, somehow machines in the creative world are
- 1:20Speaker 1: especially uncanny, even frightening, because poetry and music and humor
- 1:24Speaker 1: are supposed to be the things that define our humanity,
- 1:27Speaker 1: aren't they. In this episode, we look at how AI
- 1:30Speaker 1: is being used in the creative arts, and in doing so,
- 1:33Speaker 1: we understand a lot more about how this often intimidating
- 1:36Speaker 1: technology actually works. I'm Moza Lush and welcome to Sleepwalkers. Hi,
- 1:56Speaker 1: Hi Karen So did Oscar's poems spring your delight? I
- 2:01Speaker 1: would have preferred it in your British act? Then I think, well,
- 2:04Speaker 1: I can't blame you for that. It did remind me
- 2:06Speaker 1: of Um. My friend is in a band called the
- 2:08Speaker 1: x Ambassadors and they partnered with this producer called Alex
- 2:12Speaker 1: the Kid who was asked by IBM to make a
- 2:15Speaker 1: song using Watson. And the song is not bad. The
- 2:17Speaker 1: song is not bad. Um, the sounds good. But basically,
- 2:21Speaker 1: the way in which they used Watson was that they
- 2:24Speaker 1: crunched the twenty six thousand songs from the top one charts.
- 2:30Speaker 1: It's from over what time period, like over a few
- 2:32Speaker 1: years presume, I don't know, But the point is that
- 2:36Speaker 1: they used them to discover patterns in the songs. What
- 2:39Speaker 1: makes the top hundred song. Basically that's right, and then
- 2:42Speaker 1: that's reproduce it right, which is interesting because I think
- 2:44Speaker 1: anybody could kind of tell you what makes the top
- 2:46Speaker 1: on hundred song right. They're called earworms. But when you
- 2:49Speaker 1: think about a data set of twenty six thousand, like,
- 2:52Speaker 1: no human being can listen to that many songs and
- 2:54Speaker 1: do any productive after they hear it. So in this episode,
- 2:57Speaker 1: we're going to look at AI and art and different
- 3:00Speaker 1: kinds of fields to really understand how computers crunched data
- 3:03Speaker 1: to crack open this creative code. But first I want
- 3:06Speaker 1: to go back to the algorithm who wrote the poem
- 3:09Speaker 1: we heard at the beginning of the episode, because that
- 3:12Speaker 1: algorithm also wrote the film that Steven Spielberg of the
- 3:15Speaker 1: machine learning world, and the algorithm is called Benjamin. So
- 3:18Speaker 1: we're going to meet Benjamin and a few people not
- 3:22Speaker 1: named Benjamin. I was a speech trator for John Kerry,
- 3:27Speaker 1: Tim Gaitner, and Barack Obama, um, not in that order,
- 3:30Speaker 1: And I'm essentially a ghost writer and a photographer who
- 3:32Speaker 1: learned to code one giant Frankenstein monster. That's Ross Goodwin
- 3:37Speaker 1: and that Frankenstein monster it goes by Benjamin and it's
- 3:41Speaker 1: the work of Ross Goodwin and Oscar Sharp, who you'll
- 3:44Speaker 1: remember from that poetry reading, but neither of them actually
- 3:47Speaker 1: named Benjamin. Well, they named itself, or rather, there was
- 3:51Speaker 1: a piece of paper that came out of it that
- 3:53Speaker 1: said my name is Benjamin on it. I read that
- 3:56Speaker 1: in response to a question that was put to it,
- 3:58Speaker 1: and a room full of people went, oh, a program
- 4:02Speaker 1: that names itself is rather uncanny. And Oscar had been
- 4:06Speaker 1: chasing that uncanny nous ever since he was at n
- 4:08Speaker 1: y U for graduate school. Whenever I met anyone who
- 4:11Speaker 1: good program, I would grab them by the pels and
- 4:14Speaker 1: yell into the face. Can you can you build something
- 4:16Speaker 1: that can write like people talk in some way? And
- 4:19Speaker 1: one day in class Oscar notices there's this sneakerhead and
- 4:22Speaker 1: he's sitting on his laptop and his laptop is writing
- 4:25Speaker 1: without him touching it. And I'm like, oh, so we go,
- 4:30Speaker 1: So we go for coffee, and that gut coffee was.
- 4:33Speaker 1: It was a lengthy coffee. We're still having that coffee.
- 4:35Speaker 1: We're still having such a cold coffee. By now, you
- 4:39Speaker 1: might not believe it if it happened in a film,
- 4:42Speaker 1: but Oscar had stumbled on exactly the person he was
- 4:44Speaker 1: looking for. Oscar came to me and he said, I
- 4:47Speaker 1: want to make a movie from a computer generated screenplay.
- 4:50Speaker 1: And I said, you know, of course that sounds amazing.
- 4:52Speaker 1: Let's do it. But let's figure out how we're going
- 4:54Speaker 1: to generate the screenplay, because that's a nuanced process with
- 4:57Speaker 1: lots of stabs, and we need to consider like every
- 4:59Speaker 1: part it. So Oscar volunteered himself to teach me all
- 5:03Speaker 1: the things about storytelling and narrative and filmmaking. He turned
- 5:07Speaker 1: me onto like Vladimir prop Joseph campbell Um, all these
- 5:11Speaker 1: theories of storytelling, and so they begin to experiment. I
- 5:16Speaker 1: tried a bunch of prototypes that used like various structures
- 5:19Speaker 1: that had been postulated by these theorists over time, and
- 5:23Speaker 1: the output was not interesting. Despite following the rules laid
- 5:27Speaker 1: out by narrative theorists, Ross couldn't get anything good just
- 5:31Speaker 1: telling his programs what a story should contain. So a
- 5:35Speaker 1: year passes, Oscar moves to l A and when I
- 5:39Speaker 1: get this email from Ross and it's the results in
- 5:42Speaker 1: one of those experiments that he wants me to read,
- 5:44Speaker 1: and read it he did. Rossity mailed the poem from
- 5:47Speaker 1: the beginning of the episode, the room is blown away
- 5:50Speaker 1: from the door and the stones are beginning to shine.
- 5:53Speaker 1: I immediately was like, oh my god, I don't know
- 5:55Speaker 1: how he's doing this. But he said, I don't know
- 5:57Speaker 1: what technology you're using right now, but can we it
- 6:00Speaker 1: for screenplay? And so they did, and not just a screenplay,
- 6:03Speaker 1: they actually produced a short film called Sunspring and they
- 6:06Speaker 1: even got Thomas middle Ditch, the lead on Hbos Silaken Valley,
- 6:10Speaker 1: to star in it. Principle is completely constructed. Of the
- 6:15Speaker 1: same time, it's all about you. To be true, you
- 6:18Speaker 1: didn't even watch the movie with the rest of the base.
- 6:20Speaker 1: I don't know, I don't care. I know it's a
- 6:22Speaker 1: consequence whatever you need to know about the presence of
- 6:26Speaker 1: the story. I'm a little bit of a boy on
- 6:29Speaker 1: the floor. So what do you think, Carol? It kind
- 6:32Speaker 1: of reminds me of when my parents used to take
- 6:34Speaker 1: me to like a bad production of Macbeth or as
- 6:37Speaker 1: you like it. Traumatic. You're there and you're seven or eight,
- 6:42Speaker 1: and you want to understand what's going on, and so
- 6:45Speaker 1: you kind of pay as close attention as you possibly
- 6:47Speaker 1: can to what the actors are doing because you have
- 6:50Speaker 1: no idea what the dialogue means. Yeah, I mean that
- 6:53Speaker 1: to me is quite impressive because a machine can create
- 6:57Speaker 1: something which has enough of the elements in common the
- 7:00Speaker 1: film that we can talk about a real film. You
- 7:02Speaker 1: can't say it's not a film absolutely. Of course. What's
- 7:05Speaker 1: different is it didn't take Benjamin very long at all
- 7:07Speaker 1: to make it. Once you press the button fraction of
- 7:10Speaker 1: a second there was a couple of seconds perpase, maybe
- 7:12Speaker 1: maybe a couple of seconds total, actually a fraction of
- 7:14Speaker 1: a second per page. That's right. After months of agonizing
- 7:17Speaker 1: over centuries of storytelling theory, the final output only took
- 7:21Speaker 1: a couple of seconds. So what was Ross's breakthrough? To
- 7:25Speaker 1: understand we turned to one of the most famous AI
- 7:28Speaker 1: scientists in the world, Sebastian Throne. Recently, something magical had
- 7:32Speaker 1: happened recently. The feat has discovered was called machine learning.
- 7:38Speaker 1: With AI, computers can now find their own rules. They
- 7:41Speaker 1: are called neural networks. They're comprised of hundreds of millions
- 7:45Speaker 1: of little vase sample processing units, and those units are
- 7:48Speaker 1: modeled after what a neurons do in our physical brains.
- 7:52Speaker 1: You just give them examples, very much like the way
- 7:54Speaker 1: we we waste children. We don't give our children rules
- 7:58Speaker 1: for every contingency in life. In the first eight years
- 8:01Speaker 1: of education. We let them learn, They experience the world,
- 8:05Speaker 1: and they loan behold. They make their own rules. And
- 8:07Speaker 1: we are now in the world where computers can do
- 8:09Speaker 1: the same thing. And this means machine learning can be
- 8:12Speaker 1: used in all kinds of different fields. Sebastian himself applied
- 8:16Speaker 1: the technology at Google, where he led the initial development
- 8:19Speaker 1: of their self driving car. When you want to read
- 8:21Speaker 1: a book, a book on like what the car should
- 8:24Speaker 1: do in every situation, that rule book is really complicated
- 8:27Speaker 1: and it can promise you no matter how many years
- 8:29Speaker 1: you spent writing it, it's not gonna work. But when
- 8:33Speaker 1: you give the machine the ability to learn its own rules,
- 8:36Speaker 1: it is actually able to surpass how people can drive.
- 8:41Speaker 1: We'll hear more from Sebastian later, but machine learning mL
- 8:45Speaker 1: is the engine that drives almost all of the excitement
- 8:48Speaker 1: about AI today, from identifying targets on the battlefield to
- 8:52Speaker 1: understanding genetic diseases. And it's also what allowed Ross and
- 8:56Speaker 1: Oscar to create a usable movie script. Rather than laying
- 9:00Speaker 1: down storytelling rules, they simply showed Benjamin hundreds of examples
- 9:04Speaker 1: and the algorithm found patterns and learned for itself more sleepwalkers.
- 9:12Speaker 1: After the break, you're like, oh, did we essentially we
- 9:23Speaker 1: teach this algorithm anything else about screenplay other than just
- 9:27Speaker 1: putting in a bunch of screenplays, right, And that's the
- 9:30Speaker 1: way that machine learning works. What is happening in a
- 9:33Speaker 1: deep learning algorithm of this kind is it's building an
- 9:36Speaker 1: extraordinarily complicated mathematical formula by reading all of this stuff
- 9:41Speaker 1: over and over again, like the auto complete on your phone.
- 9:43Speaker 1: The neural that is actually sampling from a probability distribution
- 9:46Speaker 1: of which letters, bass or punctuation become next. So the
- 9:49Speaker 1: script for sun Spring was essentially the most mathematically probable
- 9:53Speaker 1: Sci Fi script except Ross and Oscar did have one
- 9:56Speaker 1: important lever of creative control. The other of parameters that
- 10:01Speaker 1: you're probably wondering about, there's one called the temperature is
- 10:04Speaker 1: the riskiness of those next letter predictions. What Ross is
- 10:08Speaker 1: describing is almost like a dial for creativity. Turn it
- 10:12Speaker 1: up to a really high temperature, and the neural net
- 10:14Speaker 1: is going to be extra creative and start making up words,
- 10:17Speaker 1: babbling at a very high temperature. It's essentially drunk. Low temperature,
- 10:22Speaker 1: it's going to be very repetitive and possible even begin
- 10:25Speaker 1: to plagiarize its source material. So it'll be very repetitive.
- 10:28Speaker 1: It'll be like the streets and the streets, and the
- 10:30Speaker 1: streets and the streets. It's essentially went working for network television. Yeah, exactly.
- 10:34Speaker 1: So we wanted it to be sort of in the middle.
- 10:37Speaker 1: In the middle is where we found the best output
- 10:39Speaker 1: and the most I think usable output, and Sunspring was born.
- 10:45Speaker 1: So Benjamin Ross and Oscar right together now they write
- 10:49Speaker 1: poetry and movies and sometimes what Benjamin spits out is good.
- 10:53Speaker 1: Often they have to sift through it to find the
- 10:55Speaker 1: best stuff. But he's prolific and he never ever suffers
- 10:59Speaker 1: from writers. Look, so Kara was telling us earlier about
- 11:04Speaker 1: Alex the kidd and using AI to make music, and
- 11:07Speaker 1: that's something I want to understand a bit more about.
- 11:09Speaker 1: So Julian went on a little bit of an expedition. Yes,
- 11:13Speaker 1: I did. I've been seeing a lot of articles lately
- 11:15Speaker 1: about AI and the arts, and I've been pretty curious
- 11:18Speaker 1: about music specifically. We might take it for granted, but
- 11:20Speaker 1: music is this primal emotional thing that's been with us forever.
- 11:24Speaker 1: It might even predate language. But now Warner Music Group
- 11:27Speaker 1: made history in April two nineteen, is the first major
- 11:30Speaker 1: label to sign an AI to a record deal. Yeah,
- 11:33Speaker 1: they signed this bot called Endel, which makes ambient noises
- 11:37Speaker 1: based on where you are and what the weather is
- 11:39Speaker 1: and what time of day it is. When I think
- 11:41Speaker 1: of this kind of music, I think of those Spotify
- 11:44Speaker 1: playlists like Peaceful Piano and Blissed Out Dinner Party, which
- 11:48Speaker 1: would become extremely popular. It's not the same thing as Beyonce, No,
- 11:52Speaker 1: definitely not. But Warner Music Group signed Endl to generate
- 11:56Speaker 1: twenty albums of ambient music. And now that we live
- 11:59Speaker 1: in a world where aies can get record deals, what
- 12:03Speaker 1: does this mean for artists? What does this mean for
- 12:05Speaker 1: even just music as we know it? Well, in my
- 12:09Speaker 1: quest to find out, I visited this company called Amper.
- 12:12Speaker 1: My name is Drew Silverstein. I am the co founder
- 12:15Speaker 1: and CEO of Amper Music. Amper is an AI music
- 12:18Speaker 1: company that Drew says will enable anyone to create music.
- 12:22Speaker 1: In fact, the only things you need to know are
- 12:24Speaker 1: the genre of music you want to create, the mood
- 12:27Speaker 1: you'd like to convey, and the length of your piece
- 12:29Speaker 1: of music. That's all you know. You can create a
- 12:31Speaker 1: brand new, unique piece of music in a matter of seconds.
- 12:34Speaker 1: So the big question is should musicians worry about computers
- 12:39Speaker 1: taking their job? Well, let's try it and see. So
- 12:43Speaker 1: what do you want to do? Cinematic, documentary, folk cinematic cinematic,
- 12:48Speaker 1: minimal percussion or quirky percussion. It's rendering a song right now.
- 12:54Speaker 1: And here we go. We've got something. I'm were deep
- 13:00Speaker 1: in the forest of Nicaragua. There's a breed of jaguar.
- 13:05Speaker 1: You might have heard of it. It's called the take
- 13:08Speaker 1: a Killer panther. Look, here's the thing. I don't know
- 13:12Speaker 1: what the difference between that and music is. I really don't. Yeah,
- 13:17Speaker 1: so you're woud Yeah, all right, So there's that. And
- 13:21Speaker 1: this isn't the only AI music app out there. Another
- 13:24Speaker 1: major player is called Magenta, and big surprise, they're at Google.
- 13:28Speaker 1: Magenta are using AI to create a ton of new tools.
- 13:31Speaker 1: From a piano genie that makes it impossible to play
- 13:34Speaker 1: bad notes to something that can generate drum loops, or
- 13:37Speaker 1: something that can even play piano duets with you. You
- 13:40Speaker 1: can even translate raw audio to a piano score. Raw audio,
- 13:44Speaker 1: like if I play just something raw on the piano,
- 13:47Speaker 1: raw audio like oh, literal raw audio, literal raw audio.
- 13:54Speaker 1: And Magenta has also trained a neural network just like
- 13:57Speaker 1: Ross and Oscar, only instead of sci fi scripts, they
- 14:00Speaker 1: trained on over four hundred performances by skilled pianists. They
- 14:04Speaker 1: fed it into the neural network and let me play
- 14:07Speaker 1: one of the piano experts. First. This is a real
- 14:09Speaker 1: piano player, all right, so nice? Right? Yeah, okay, ready
- 14:17Speaker 1: for the AI. What that's a computer that was all
- 14:28Speaker 1: a computer. I didn't ever play a human one. That's
- 14:30Speaker 1: a computer that was trained by a human playing piano.
- 14:34Speaker 1: And then how do you make a computer come up
- 14:37Speaker 1: with that? Right? So even though it's not a screenplay,
- 14:39Speaker 1: it's still data that you can feed a neural network
- 14:41Speaker 1: with to find patterns. And in this case, Magenta used
- 14:44Speaker 1: a data set from the Yamahai Piano competition. So human
- 14:47Speaker 1: pianists played on these digital keyboards which recorded the nuances
- 14:50Speaker 1: of their performance, like how long they hit notes, and
- 14:52Speaker 1: it recorded all that information into a digital score that
- 14:55Speaker 1: a computer could interpret. And we've actually had that technology
- 14:57Speaker 1: for a while now, it's called MIDI. But training and
- 15:00Speaker 1: on network on the data is new. See. The thing
- 15:02Speaker 1: that I come back to is that a computer doesn't
- 15:05Speaker 1: know it's playing music, so much of watching a musical
- 15:09Speaker 1: performance is knowing that this is coming from someone who
- 15:13Speaker 1: is emoting. Right, Yeah, there's actually there's an emotional communication happening, right,
- 15:18Speaker 1: that's right. I do think though the future is not
- 15:22Speaker 1: rejecting this. It's better to imagine what would Stravinsky have
- 15:26Speaker 1: done with this kind of technology, because Stravinsky is still
- 15:29Speaker 1: a musical genius. Right, Yeah, Definitely it's cool to listen
- 15:42Speaker 1: to those musical examples of machine learning because you can
- 15:45Speaker 1: really hear how the algorithm is reinterpreting existing material. Of course,
- 15:51Speaker 1: listening to the output is one thing. Tasting it is
- 15:54Speaker 1: quite another. The problem was that somebody had told me
- 15:58Speaker 1: that they had made the recipe for Stan, that it
- 16:00Speaker 1: was good, and what it was as a recipe called
- 16:03Speaker 1: chocolate baked and serves. That's Janelle Shane. She's a research
- 16:08Speaker 1: scientist and the author of a blog called AI Weirdness.
- 16:11Speaker 1: She's talking about a recipe written by Ai that she
- 16:14Speaker 1: actually cooked and eight. It starts out as a perfectly ordinary,
- 16:19Speaker 1: flowerless chocolate brownie all the way until the very last ingredient,
- 16:24Speaker 1: which is a cup of horse badish. I knew I
- 16:26Speaker 1: was in trouble when I opened the oven door and
- 16:28Speaker 1: my eyes just started watering. It was yeah, it was terrible.
- 16:34Speaker 1: On her blog, Jenelle experiments with putting AI to a
- 16:37Speaker 1: range of tasks, from writing new pickup lines to naming
- 16:40Speaker 1: Halloween costumes, and often her experiments with machine learning are
- 16:44Speaker 1: pretty revealing about us. It plays into this thought experiment,
- 16:49Speaker 1: what would an alien think of our world? It takes
- 16:52Speaker 1: something that's very ordinary and mixes it up into this
- 16:57Speaker 1: thing that sounds like the original, but the meaning has
- 17:01Speaker 1: been completely changed. Chopped whipping cream may be an ingredient
- 17:05Speaker 1: and fold water, enrolled it into cubes, or spread the
- 17:08Speaker 1: butter in the refrigerator. That's another direction that came up with.
- 17:12Speaker 1: Remember Ross and Oscar playing with the creativity setting for
- 17:15Speaker 1: their scripts. Janelle plays with herbot's temperature too, so I
- 17:19Speaker 1: can turn it up and the neural net may choose
- 17:22Speaker 1: its second best or third best guess as to what
- 17:25Speaker 1: letter comes next. And if I turn the creativity all
- 17:28Speaker 1: the way down, then everything maybe something like the the
- 17:32Speaker 1: the or recipes may be just you know, one teaspoon
- 17:36Speaker 1: of vanilla over and over and over again, because that's
- 17:39Speaker 1: just a very likely ingredient. It's really interesting with the
- 17:44Speaker 1: recipes to turn down the creativity and see what it
- 17:48Speaker 1: comes up with as the most quintessential recipes. At the
- 17:52Speaker 1: lowest setting, you may not get hole Strandish brownies, but
- 17:55Speaker 1: you do get a clear picture of what we eat
- 17:57Speaker 1: and who we are. I look at what kinds of
- 18:00Speaker 1: recipe titles that comes up with. There are things like
- 18:03Speaker 1: chocolate chicken chicken cake, and another one that's chocolate chocolate
- 18:08Speaker 1: chocolate chocolate cake. And there was a lot of cheese
- 18:10Speaker 1: in these recipes too, so it's kind of revealing about
- 18:14Speaker 1: what sorts of things we cook with. Then we like chocolate,
- 18:18Speaker 1: cheese and chicken apparently. But then I did the same
- 18:22Speaker 1: experiment with recipes from Bone Appetite, and then the most
- 18:26Speaker 1: common ingredients that kept using were cilantro and pomegranate juice.
- 18:32Speaker 1: So these algorithms essentially hold up a mirror to the
- 18:35Speaker 1: data sets that we give them. They do, yeah, they
- 18:37Speaker 1: reflect the data sets back to us in really weird ways,
- 18:42Speaker 1: and they can absolutely pick up whatever bias there is
- 18:45Speaker 1: in a input data set. And I think what we're
- 18:48Speaker 1: discovering is just how prevalent that bias is and how
- 18:52Speaker 1: easy it is for neural networks to latch onto that
- 18:55Speaker 1: bias and copy it as a handy tool toward copying
- 18:59Speaker 1: whatever where the humans are doing. They say that the
- 19:02Speaker 1: way to a person's heart is through their stomach. But
- 19:05Speaker 1: Janelle didn't stop at chocolate, bakes and surfs. She's also
- 19:09Speaker 1: turned AI onto some more direct roots. I really liked
- 19:12Speaker 1: the pickup lines. And there are all these puns and
- 19:15Speaker 1: all this wordplay that it didn't have any way to
- 19:17Speaker 1: grab hold of and figure out how to use. But
- 19:20Speaker 1: I think what it produced this sort of charming surrealism
- 19:26Speaker 1: and kind of garble nonsensical. I think it's an improvement
- 19:30Speaker 1: on every single one of the originals. My very favorite
- 19:34Speaker 1: one is you look like a thing and I love you.
- 19:37Speaker 1: You are so beautiful that you make me feel better
- 19:39Speaker 1: to see you. Or you must be a tringle because
- 19:42Speaker 1: you're the only thing here. Are you a camera? Because
- 19:46Speaker 1: I want to see the most beautiful than you. Yeah,
- 19:49Speaker 1: I'll definitely lie with you. No one's have a used
- 19:54Speaker 1: real pickup line on me? Use one on you know?
- 19:57Speaker 1: Do I look like someone who would receive a pickup Well,
- 20:01Speaker 1: here's one of them. I don't know you. That's good
- 20:05Speaker 1: a lot of girls are into that. Are you a candle?
- 20:07Speaker 1: Because you're so hard of the looks with you. So
- 20:12Speaker 1: in effect, what the algorithm is doing is highlighting patterns
- 20:15Speaker 1: in the data. I mean, there's sound structural like pickup
- 20:18Speaker 1: lines about the words themselves don't make any sense. The
- 20:21Speaker 1: machines are reflecting their creators and spitting back something which
- 20:24Speaker 1: resembles pick up lines and makes us think a little
- 20:27Speaker 1: bit more carefully about what a pickup line is. And
- 20:31Speaker 1: while training Benjamin Ross and Oscar found the same thing
- 20:34Speaker 1: as the algorithm learned patterns revealed bias present in our cinema.
- 20:40Speaker 1: When you train an algorithm like Benjamin on millions and
- 20:43Speaker 1: millions in this case of synopsis from the Internet of
- 20:45Speaker 1: the movies, the synopsis that come out have certain patterns
- 20:49Speaker 1: in them. For example, they mentioned men full times more
- 20:51Speaker 1: often than they mentioned women. But you you learn other
- 20:54Speaker 1: things about it than that. You learned that the most
- 20:57Speaker 1: common phrase in the in the output is a young
- 20:59Speaker 1: man in a small town. So what does a filmmaker
- 21:02Speaker 1: like Oscar learn from this? I used to call this
- 21:04Speaker 1: project the average movie projects. And the reason I called
- 21:06Speaker 1: it that is the theory was for me, if you
- 21:09Speaker 1: could make the right kind of algorithm that the movie
- 21:11Speaker 1: that you would make that would be the theoretically perfect
- 21:14Speaker 1: movie would also be the most boring movie ever made,
- 21:17Speaker 1: and that it would. It would it would be by
- 21:19Speaker 1: definition all of the things that were the most clich
- 21:21Speaker 1: because that's what cliche means, is the thing that that
- 21:23Speaker 1: you can rely on to work. And why do that?
- 21:26Speaker 1: Because the thing I'm most interested in is doing the
- 21:28Speaker 1: thing that we haven't done yet. I want to move
- 21:30Speaker 1: the form forward. Seeing all of these biases and assumptions
- 21:33Speaker 1: that are baked into our movies and our snacks doesn't
- 21:36Speaker 1: mean we're doomed to repeat them. In fact, the awareness
- 21:39Speaker 1: can be liberating. That's what's helped me, I think, is
- 21:42Speaker 1: seeing Benjamin's capacity to show me more directly what it
- 21:46Speaker 1: is that are our patents, are our habits, and then
- 21:49Speaker 1: I can ask more easily how to move forward from that.
- 21:53Speaker 1: We'll get there after the break. So we've heard about
- 22:05Speaker 1: Janelle Shane using AI to reveal bias, and Ross and
- 22:08Speaker 1: Oscar using it to help them think more creatively about
- 22:10Speaker 1: filmmaking as well as how it can be applied to music.
- 22:14Speaker 1: And that's the great promise of AI. We may worry
- 22:17Speaker 1: about replacing jobs, but it can augment our lives in
- 22:20Speaker 1: so many ways. At least that's how Sebastian Throne sees it.
- 22:24Speaker 1: I would say the term AI is a bit deceptive
- 22:27Speaker 1: because it sets up computers to be on equal power
- 22:29Speaker 1: with people. I see it to be stronger where we
- 22:33Speaker 1: are weak, and weaker where you're strong. It's not a
- 22:36Speaker 1: technology that will replace us, as it's not really empowers
- 22:40Speaker 1: but what might that empowerment look like beyond bias detection
- 22:44Speaker 1: and piano playing well. In seventeen, Sebastian published a paper
- 22:48Speaker 1: in Nature on using AI to diagnose skin cancer using
- 22:52Speaker 1: just an iPhone. So in medicine, you can think of
- 22:56Speaker 1: your iPhone that can find skin cancer as turning regular
- 23:01Speaker 1: physicians or anybody in the world into an expert on
- 23:04Speaker 1: day one, because now they have the superpower to be
- 23:07Speaker 1: able to distinguish something that previously would have taken tens
- 23:10Speaker 1: of years to learn. The same is true for the
- 23:13Speaker 1: self driving car. Now children can drive and and and
- 23:16Speaker 1: blind people can drive, head blind people drive around and
- 23:18Speaker 1: self driving cars. So for me, the real opportunities to
- 23:22Speaker 1: use the eye to extract the knowledge from some human
- 23:24Speaker 1: experts that are well trained and transpose this knowledge to
- 23:27Speaker 1: other brains that people not so well trained. This is
- 23:30Speaker 1: what I personally find so fascinating about AI and a
- 23:33Speaker 1: big reason I wanted to do Sleepwalkers. The same technology
- 23:37Speaker 1: underlies rossan Oscar's films, Janelle's recipes and self driving cars
- 23:42Speaker 1: and cancer diagnostics and so much more. The training for
- 23:47Speaker 1: skin cancer detection or cancer detection radeology and the training
- 23:50Speaker 1: for the self dime car amazingly similar. In both cases,
- 23:55Speaker 1: what you do is you compile the data set typically
- 23:57Speaker 1: hundreds of thousands up to hundreds of millions of images.
- 24:01Speaker 1: In skin cancer, we use biopsies. We had a database
- 24:04Speaker 1: of a hundred twenty nine thousand images that a lab
- 24:07Speaker 1: had biopsied and provided. In the self driving car, you
- 24:11Speaker 1: could be as easy as having a human driver provide
- 24:13Speaker 1: inputs with their student veel and the and the break
- 24:16Speaker 1: as to what the right thing is to do, and
- 24:19Speaker 1: then the network mimics human behavior, it mimics the diagnostics
- 24:22Speaker 1: of a physician, or it mimics the style of a driver.
- 24:25Speaker 1: The underlying albums are amazingly similar. What makes this moment
- 24:29Speaker 1: all the more interesting is that AI is in the
- 24:32Speaker 1: process of being consumerized like Sebastion said, your iPhone can
- 24:37Speaker 1: diagnose skin cancer. Self driving cars are already on the roads,
- 24:41Speaker 1: and as these tools become more and more accessible, society
- 24:44Speaker 1: will start to change. These technologies become closer and closer
- 24:48Speaker 1: to us. The fact that you carry yourself phone is
- 24:51Speaker 1: a big deal. You might not see this way, but
- 24:54Speaker 1: what it does it puts the computer seamlessly into your life.
- 24:58Speaker 1: You're texting app, your SMS is so close to you
- 25:01Speaker 1: that you can now talk to people thousands of miles
- 25:04Speaker 1: of a on a button press or on a microphone.
- 25:06Speaker 1: That makes you effectively super human without the actual physical implant.
- 25:11Speaker 1: But when it comes to AI, for now, leading edge
- 25:14Speaker 1: algorithms are off limits to those of us who can't
- 25:17Speaker 1: code or who don't have the means to learn. Oscar
- 25:20Speaker 1: wouldn't have been able to make Sunspring without Rosses technical expertise,
- 25:24Speaker 1: but that's all starting to change, Julian, you spoke with
- 25:28Speaker 1: somebody making AI more accessible. Yeah, while we were looking
- 25:31Speaker 1: into AI and music, we came across Runway m L
- 25:34Speaker 1: their lab based in Bushwick, and they feel strongly about
- 25:37Speaker 1: letting more people into work with AI Creatively. I spoke
- 25:41Speaker 1: with Christo bal Valence Weela, the co founder of Runway,
- 25:43Speaker 1: which is basically like the Adobe Creative Suite for AI.
- 25:46Speaker 1: So think of a program that looks like Photoshop. They're
- 25:49Speaker 1: adding tons of different AI models to the Runway app,
- 25:51Speaker 1: where instead of having to know how to code, you
- 25:54Speaker 1: can just manipulate some sliders and dials and still have
- 25:57Speaker 1: a I generate something. If we really think of this
- 26:00Speaker 1: the game changing technology that will impact us for like
- 26:02Speaker 1: years to come, we need to have more people from
- 26:05Speaker 1: different backgrounds and disciplines jumping to that discussion and proposing
- 26:09Speaker 1: ways of looking at algorithms that those researchers and those
- 26:11Speaker 1: scientists are not thinking of. This is gonna impact us,
- 26:14Speaker 1: and it's going to change the way we see not
- 26:15Speaker 1: just the world, but ourselves. Not just ourselves, but how
- 26:18Speaker 1: we think about our creativity. And the list of things
- 26:21Speaker 1: that Runway can help you do is frankly crazy. Will
- 26:24Speaker 1: post some on our Instagram at Sleepwalkers podcast, but just
- 26:28Speaker 1: one example, you can take video of anyone adds and
- 26:31Speaker 1: have their body copy the poses that you make in
- 26:34Speaker 1: your webcam. So Krystoval actually tweeted one of him controlling
- 26:37Speaker 1: Stephen Colbert's body on The Late Show just with his
- 26:40Speaker 1: webcam moving his arms. Stephen moves his arms. It's nuts, right,
- 26:44Speaker 1: And imagine that kind of technology in the hands of artists.
- 26:47Speaker 1: Start thinking about them as not like something that's gonna
- 26:50Speaker 1: destroy our creativity or gonna replace writers an artists or whatever.
- 26:53Speaker 1: This is gonna be a typewriter, This is gonna be
- 26:55Speaker 1: a paintbrush, and people will start building and using it
- 26:58Speaker 1: to understand their own creativity in a new way. Of course,
- 27:02Speaker 1: as always, it's up to us to make sure we
- 27:05Speaker 1: use these new tools for good. If I build a shovel, okay,
- 27:09Speaker 1: and you decide to go to the beach and digging sand,
- 27:12Speaker 1: you're biased. You're digging in sand and guess what You're
- 27:15Speaker 1: shoveled only to an up sand. The same as true
- 27:17Speaker 1: for AI. If you give it a certain type of
- 27:19Speaker 1: data set, I can promise you whatever I get out
- 27:21Speaker 1: reflects the data you're put in. It's up to us,
- 27:24Speaker 1: the people, to make responsible decisions. And as we want
- 27:28Speaker 1: to create equal opportunity and evadicate certain biases society that
- 27:31Speaker 1: exists today, is up to us to do it, and
- 27:34Speaker 1: I promise you, if you work hard on this, technologies
- 27:37Speaker 1: will reflect that. But even Sebastian, one of Silicon Valleys
- 27:41Speaker 1: Great Optimists, recognizes the risks all technologies can harm people.
- 27:47Speaker 1: In fact, technologies can be abused to harm people, Like
- 27:50Speaker 1: my kitchen knife, which serves me a great purpose every
- 27:53Speaker 1: time I've guests over in shopping. My produce can also
- 27:56Speaker 1: be abused to harm people. In the next episode of Sleepwalkers,
- 28:01Speaker 1: we dive deep into the ability of algorithms to cause harm.
- 28:04Speaker 1: We traveled from China and the social credit system to
- 28:08Speaker 1: a parole board in New York and we speak with
- 28:10Speaker 1: people building technology they believe will make us safer. I'm
- 28:15Speaker 1: as Vlachen, see you next time. Sleepwalkers is a production
- 28:31Speaker 1: of our heart Radio and unusual productions. There's so much
- 28:34Speaker 1: we don't have time for in our episodes, but that
- 28:37Speaker 1: we'd love to share with you. So for the latest
- 28:39Speaker 1: AI news, live interviews, and behind the scenes footage, find
- 28:43Speaker 1: us on Instagram, at Sleepwalker's podcast or at sleepwalkers podcast
- 28:47Speaker 1: dot com. Special thanks on this episode to paw Suris,
- 28:50Speaker 1: who introduced us to Oscar Ross and Benjamin and to
- 28:54Speaker 1: artificial intelligence, which composed over half of music in this episode.
- 28:58Speaker 1: Could you tell which was which Sleepwalkers is hosted by
- 29:02Speaker 1: me Ozveloshin and co hosted by me Kara Price. Were
- 29:05Speaker 1: produced by Julian Weller with help from Jacopo Penzo and
- 29:08Speaker 1: Taylor Chacog. Mixing by Tristan McNeil and Julian Weller. Our
- 29:12Speaker 1: story editor is Matthew Riddle. Recording assistance. This episode from
- 29:15Speaker 1: Joanne de Luna. Sleepwalkers is executive produced by me Ozveloshin
- 29:20Speaker 1: and Mangesh hatt Together. For more podcasts from my heart Radio,
- 29:23Speaker 1: visit the i heart Radio app, Apple Podcasts, or wherever
- 29:26Speaker 1: you listen to your favorite shows,