WEBVTT - Using AI for Creative Work

0:00:15.356 --> 0:00:15.796
<v Speaker 1>Pushkin.

0:00:24.036 --> 0:00:26.436
<v Speaker 2>A few weeks ago, I went to Chicago to interview

0:00:26.516 --> 0:00:30.636
<v Speaker 2>two people on stage about creative work they've done using

0:00:30.756 --> 0:00:35.036
<v Speaker 2>artificial intelligence. One of the people was Stephen marsh He's

0:00:35.076 --> 0:00:39.796
<v Speaker 2>a writer. He's done nonfiction books, novels, magazine articles, and

0:00:40.076 --> 0:00:43.196
<v Speaker 2>earlier this year, he used AI to help him write

0:00:43.276 --> 0:00:46.556
<v Speaker 2>a short novel called Death of an Author. That book,

0:00:46.596 --> 0:00:50.036
<v Speaker 2>by the way, was published in audio form by Pushkin Industries,

0:00:50.076 --> 0:00:54.556
<v Speaker 2>the same company that publishes this podcast. The other person

0:00:54.636 --> 0:00:58.836
<v Speaker 2>on stage with us was Lucas Contor. Lucas is a composer.

0:00:59.196 --> 0:01:01.756
<v Speaker 2>Among other things, He's won a couple Emmys for his

0:01:01.876 --> 0:01:03.036
<v Speaker 2>work scoring.

0:01:02.676 --> 0:01:03.956
<v Speaker 1>The Olympics for NBC.

0:01:04.756 --> 0:01:07.076
<v Speaker 2>He co produced a Lord song that was in one

0:01:07.116 --> 0:01:10.356
<v Speaker 2>of the Hunger Games movies. And the reason he was

0:01:10.396 --> 0:01:13.676
<v Speaker 2>there talking with us Lucas used AI to help him

0:01:13.676 --> 0:01:18.156
<v Speaker 2>finish Schubert's unfinished symphony. It was a really interesting conversation

0:01:18.636 --> 0:01:20.276
<v Speaker 2>and I thought it would make a great episode of

0:01:20.276 --> 0:01:21.356
<v Speaker 2>What's Your Problem?

0:01:21.396 --> 0:01:24.956
<v Speaker 1>So here it is now, please join me and welcome

0:01:25.076 --> 0:01:25.596
<v Speaker 1>our panel.

0:01:31.076 --> 0:01:34.356
<v Speaker 2>So let's do a thing you're never supposed to do

0:01:34.636 --> 0:01:41.236
<v Speaker 2>in narrative. Let's answer the question right at the beginning. So, uh,

0:01:41.796 --> 0:01:46.636
<v Speaker 2>the sort of headline question for this panel is will

0:01:46.676 --> 0:01:50.116
<v Speaker 2>AI kill creativity? I want to ask both of you.

0:01:51.276 --> 0:01:53.916
<v Speaker 2>I want you to answer in one word at the

0:01:53.956 --> 0:01:56.836
<v Speaker 2>same time, on the count of three. It's gonna be

0:01:56.956 --> 0:01:59.196
<v Speaker 2>it's not one two three go, You're gonna go on

0:01:59.636 --> 0:02:01.876
<v Speaker 2>what it is? One two three go?

0:02:02.156 --> 0:02:02.436
<v Speaker 1>Yeah?

0:02:02.516 --> 0:02:07.796
<v Speaker 3>Okay, will AI kill creativity? One two three No? No,

0:02:08.676 --> 0:02:12.796
<v Speaker 3>great done, let's go. Yeah, thank you, Thank you very much.

0:02:15.636 --> 0:02:18.236
<v Speaker 2>So I'm delighted to be here with both of you,

0:02:19.596 --> 0:02:23.276
<v Speaker 2>in particular because you have made things with AI. Right

0:02:23.316 --> 0:02:25.956
<v Speaker 2>there have been countless panels of people sort of waving

0:02:25.996 --> 0:02:26.796
<v Speaker 2>their hands.

0:02:26.476 --> 0:02:29.476
<v Speaker 4>About the theory of AI or the future of AI.

0:02:30.156 --> 0:02:33.556
<v Speaker 2>But I love that we're here talking about things that

0:02:33.636 --> 0:02:36.796
<v Speaker 2>you have made, creative work that you've made. And so

0:02:36.836 --> 0:02:39.276
<v Speaker 2>what I want to do is, I want to start

0:02:39.276 --> 0:02:42.076
<v Speaker 2>by talking a little bit about process. I love talking

0:02:42.076 --> 0:02:45.916
<v Speaker 2>about how people make creative things. And we'll just do

0:02:45.996 --> 0:02:48.316
<v Speaker 2>that in order, frankly, just because I want to get

0:02:48.356 --> 0:02:51.956
<v Speaker 2>into first of the book and then the symphony, and

0:02:51.996 --> 0:02:54.636
<v Speaker 2>then we can talk more generally about AI and creativity

0:02:54.676 --> 0:02:56.996
<v Speaker 2>and humanity, and then we can wave our hands in

0:02:56.996 --> 0:03:00.516
<v Speaker 2>that classic can'd wave away. So Stephen, let's start with you.

0:03:02.036 --> 0:03:05.316
<v Speaker 2>I want to read an excerpt from your book, in

0:03:05.396 --> 0:03:08.916
<v Speaker 2>part because this book that was written with Ai has

0:03:08.956 --> 0:03:13.196
<v Speaker 2>a very particular, I don't know quality to the pros.

0:03:13.396 --> 0:03:16.516
<v Speaker 2>There's a really interesting feel to the pros, and I

0:03:16.556 --> 0:03:18.556
<v Speaker 2>don't know if you'll quite get it from a paragraph,

0:03:18.636 --> 0:03:20.996
<v Speaker 2>but I want to give you something to hold on

0:03:21.036 --> 0:03:22.316
<v Speaker 2>to as we're talking about the book.

0:03:22.356 --> 0:03:25.156
<v Speaker 1>So I think I have this right.

0:03:25.196 --> 0:03:27.356
<v Speaker 2>This passage I'm going to read, it's in the first person,

0:03:27.916 --> 0:03:32.276
<v Speaker 2>and it's actually in the book, spoken by a digital avatar,

0:03:32.316 --> 0:03:35.476
<v Speaker 2>an AI avatar in the book, who is an avatar

0:03:35.796 --> 0:03:39.156
<v Speaker 2>of a dead author whose death is the title of

0:03:39.196 --> 0:03:42.876
<v Speaker 2>the book. So the passage in her voice goes like this,

0:03:45.316 --> 0:03:47.596
<v Speaker 2>I learned the limits of machines when they wanted me

0:03:47.676 --> 0:03:50.436
<v Speaker 2>to fly bombers. They were going to force me to

0:03:50.476 --> 0:03:52.436
<v Speaker 2>push a button that would end the world.

0:03:53.116 --> 0:03:53.836
<v Speaker 1>I hope you can.

0:03:53.756 --> 0:03:58.516
<v Speaker 2>Understand that my stance as a pacifist wasn't cowardice or principle,

0:03:58.716 --> 0:04:02.436
<v Speaker 2>but a confession. I could never bring myself to press

0:04:02.476 --> 0:04:07.036
<v Speaker 2>that button. Human beings cannot stop making buttons, and once

0:04:07.076 --> 0:04:10.156
<v Speaker 2>we've made them, we can't stop pushing them.

0:04:10.876 --> 0:04:12.876
<v Speaker 1>Pretty good for a machine, it really pretty good for

0:04:12.916 --> 0:04:13.396
<v Speaker 1>a machine.

0:04:13.476 --> 0:04:15.636
<v Speaker 2>Yeah, I'm gonna read that last sentence again because I

0:04:15.716 --> 0:04:17.356
<v Speaker 2>like it, and because it comes up a couple times

0:04:17.396 --> 0:04:21.476
<v Speaker 2>in the book. Human beings cannot stop making buttons, and

0:04:21.516 --> 0:04:25.356
<v Speaker 2>once we've made them, we can't stop pushing them. So

0:04:25.636 --> 0:04:29.116
<v Speaker 2>maybe Stephen, we should actually start with that sentence.

0:04:29.396 --> 0:04:29.676
<v Speaker 1>Right.

0:04:29.956 --> 0:04:33.196
<v Speaker 2>It's a great sentence, I think, I or really interesting sentence.

0:04:33.556 --> 0:04:35.356
<v Speaker 2>Sounds like a sentence a human being would write. It

0:04:35.436 --> 0:04:37.956
<v Speaker 2>ends up being important in the book sort of thematically.

0:04:38.516 --> 0:04:41.756
<v Speaker 2>How did the machine write that sentence?

0:04:42.076 --> 0:04:45.036
<v Speaker 5>Okay, let me see if I can get it exactly right.

0:04:45.396 --> 0:04:49.436
<v Speaker 5>So that was the first person from the death of

0:04:49.476 --> 0:04:52.356
<v Speaker 5>an author. So Jacob came to me in February and said,

0:04:52.396 --> 0:04:53.556
<v Speaker 5>we need to release this thing.

0:04:53.756 --> 0:04:57.276
<v Speaker 2>This is Jacob Weisberg, actually the person who runs Yeah,

0:04:57.276 --> 0:04:59.196
<v Speaker 2>pushed It. Let's actually start at the beginning of the

0:04:59.276 --> 0:05:01.716
<v Speaker 2>sure and then we'll get to that sentence. So right,

0:05:01.836 --> 0:05:05.396
<v Speaker 2>So Jacob Weisberg, who runs Pushkin, which is the company

0:05:05.476 --> 0:05:08.476
<v Speaker 2>where I make a podcast, came to you in February.

0:05:08.196 --> 0:05:11.396
<v Speaker 5>And said, can you write a book that's AI that's

0:05:11.476 --> 0:05:13.516
<v Speaker 5>generated by AI? In fact, he said, can you create

0:05:13.556 --> 0:05:16.836
<v Speaker 5>an AI author and then have that author create a book.

0:05:16.836 --> 0:05:19.316
<v Speaker 5>Now I'd been working on this for a while i'd

0:05:19.356 --> 0:05:23.116
<v Speaker 5>been working. I'd wroteen my first you know, algorithmically generated

0:05:23.156 --> 0:05:27.476
<v Speaker 5>story for Wired in twenty seventeen, which was before the Transformer,

0:05:27.796 --> 0:05:32.236
<v Speaker 5>so the Dark Ages of AI really, and so I said, yes,

0:05:32.316 --> 0:05:35.876
<v Speaker 5>I can definitely do that. It'll be about ninety five

0:05:35.916 --> 0:05:38.716
<v Speaker 5>percent computer generated. I don't want to if I want

0:05:38.756 --> 0:05:41.356
<v Speaker 5>to change heat to the character's name or something like that,

0:05:41.396 --> 0:05:43.596
<v Speaker 5>I want to be able to do that without forcing

0:05:43.636 --> 0:05:47.396
<v Speaker 5>all these iterations and so on. And basically I used

0:05:48.276 --> 0:05:54.076
<v Speaker 5>GBT four and I would use it to generate texts.

0:05:54.396 --> 0:05:58.236
<v Speaker 5>I knew from having done AI AI text before that

0:05:58.876 --> 0:06:01.596
<v Speaker 5>A is very poor at generating plots, okay, and it's

0:06:01.676 --> 0:06:03.236
<v Speaker 5>very poor at certain other tasks.

0:06:03.236 --> 0:06:05.356
<v Speaker 1>It's incredibly good at style, okay.

0:06:05.436 --> 0:06:08.796
<v Speaker 5>Right, So I would, you know, have very clear ideas

0:06:08.796 --> 0:06:11.116
<v Speaker 5>of where the narrative what's going. I'd give very specific

0:06:11.196 --> 0:06:18.116
<v Speaker 5>grammatical and syntactical commands write a paragraph with high variability,

0:06:18.316 --> 0:06:21.476
<v Speaker 5>like very very specific commands like wait, do the whole.

0:06:21.556 --> 0:06:24.076
<v Speaker 4>Give me an example in its entirety of a command.

0:06:24.156 --> 0:06:26.116
<v Speaker 5>It would be almost impossible to do because it's exactly

0:06:26.156 --> 0:06:29.316
<v Speaker 5>like do it when you've seen them for visual stuff,

0:06:29.556 --> 0:06:31.916
<v Speaker 5>where it's like they'll just to get really interesting AI

0:06:31.996 --> 0:06:37.756
<v Speaker 5>generated pictures. You often have like one hundred different references.

0:06:36.596 --> 0:06:39.436
<v Speaker 4>Like it almost impossible, but just give me something.

0:06:39.476 --> 0:06:40.036
<v Speaker 1>Give me something.

0:06:42.436 --> 0:06:48.956
<v Speaker 5>Write a hard boiled detective story paragraph with a variability

0:06:48.956 --> 0:06:57.396
<v Speaker 5>between short and long sentences and clear, elegant syntax, containing

0:06:57.396 --> 0:07:00.116
<v Speaker 5>the following information, and then you write out information it

0:07:00.116 --> 0:07:02.996
<v Speaker 5>would generate that. Then you would take that and I

0:07:02.996 --> 0:07:04.716
<v Speaker 5>would put it into a program called pseudo.

0:07:04.756 --> 0:07:07.076
<v Speaker 2>Right, and wait just before we go to the next program,

0:07:07.156 --> 0:07:11.276
<v Speaker 2>when you say containing the following information, like that.

0:07:11.116 --> 0:07:13.236
<v Speaker 1>One would be it would be like in this one.

0:07:13.436 --> 0:07:16.996
<v Speaker 5>The author says, well, that would be slightly different because

0:07:17.196 --> 0:07:20.116
<v Speaker 5>with characters, I would use a whole different set of commands.

0:07:20.516 --> 0:07:23.476
<v Speaker 5>So you know the author and here was basically a

0:07:23.516 --> 0:07:26.676
<v Speaker 5>combination of Margaret Outwood and my dead father. Because I

0:07:26.676 --> 0:07:28.476
<v Speaker 5>was writing this thing fast, so I needed to know

0:07:28.596 --> 0:07:30.996
<v Speaker 5>something that I needed to have a character that I

0:07:31.036 --> 0:07:32.676
<v Speaker 5>would automatically be interested in, and.

0:07:32.676 --> 0:07:36.716
<v Speaker 2>I should say, you're a Canadian, like basically the next

0:07:36.716 --> 0:07:37.756
<v Speaker 2>closest thing after your.

0:07:38.156 --> 0:07:42.316
<v Speaker 5>If women were alive for you, yes, right, and so uh,

0:07:42.836 --> 0:07:46.516
<v Speaker 5>and so that I would say write something like Sylvia

0:07:46.556 --> 0:07:49.596
<v Speaker 5>Plath meets Philip Roth and meets a bunch of different

0:07:49.636 --> 0:07:51.196
<v Speaker 5>other things and get hurt it.

0:07:51.356 --> 0:07:53.876
<v Speaker 2>So you're doing a very specific character. And then do

0:07:53.916 --> 0:07:56.876
<v Speaker 2>you do all of the sort of exposition or plot points,

0:07:56.916 --> 0:07:59.396
<v Speaker 2>like what what in terms of substance, what is an

0:07:59.396 --> 0:08:00.516
<v Speaker 2>example of what you might put?

0:08:00.516 --> 0:08:03.556
<v Speaker 5>Well, I would that would probably actually be mostly the machine,

0:08:03.556 --> 0:08:05.836
<v Speaker 5>but for plot details would be like she walks to

0:08:05.916 --> 0:08:07.276
<v Speaker 5>a bridge.

0:08:07.316 --> 0:08:10.436
<v Speaker 2>And but this this paragraph about like the you know,

0:08:11.196 --> 0:08:13.476
<v Speaker 2>the buttons. I wouldn't press the button, and it's like,

0:08:13.516 --> 0:08:16.516
<v Speaker 2>how do you It would be something like you know

0:08:16.556 --> 0:08:18.636
<v Speaker 2>it to be something like the character.

0:08:19.956 --> 0:08:24.996
<v Speaker 5>Reminisces about her times as a UH and and expounds

0:08:24.996 --> 0:08:31.596
<v Speaker 5>philosophically on the difference between AI and being a fighter pilot.

0:08:31.596 --> 0:08:34.716
<v Speaker 2>And or the character expounds on being a pacifist in

0:08:34.756 --> 0:08:39.036
<v Speaker 2>the military. Exactly okay, right, and so sometimes more, sometimes less.

0:08:39.436 --> 0:08:41.756
<v Speaker 2>Tried to get as little as possible, but you know

0:08:41.916 --> 0:08:45.516
<v Speaker 2>you want specificity here, like you're the more precise the command,

0:08:45.596 --> 0:08:49.116
<v Speaker 2>the better information, the precise the command, the more it's

0:08:49.156 --> 0:08:52.236
<v Speaker 2>just you writing it with the weird kind of intermediation is.

0:08:52.276 --> 0:08:55.476
<v Speaker 5>My creation, right, this is a tool which you will

0:08:55.636 --> 0:08:58.356
<v Speaker 5>I will say the same thing. So just the same

0:08:58.356 --> 0:09:01.196
<v Speaker 5>as if like this is the thing people don't understand, right,

0:09:01.276 --> 0:09:03.356
<v Speaker 5>it's like, of course this is a creative act. It's

0:09:03.356 --> 0:09:05.916
<v Speaker 5>just a different creative act, right, Like it's this is

0:09:05.956 --> 0:09:08.676
<v Speaker 5>one hundred percent me. It's just I didn't write the

0:09:08.716 --> 0:09:14.356
<v Speaker 5>words right like like like so that's like like that's weird,

0:09:14.476 --> 0:09:16.916
<v Speaker 5>Like I am, yeah, it's very weird, like I am.

0:09:16.996 --> 0:09:18.876
<v Speaker 6>Don't you didn't write the words that ended up in

0:09:18.916 --> 0:09:21.916
<v Speaker 6>the book word you weren't the words that were the

0:09:21.956 --> 0:09:23.756
<v Speaker 6>instructions to the machine to write the words that.

0:09:23.996 --> 0:09:26.836
<v Speaker 1>Well, so good as any computer that's true, any computer program.

0:09:26.876 --> 0:09:28.436
<v Speaker 2>So so okay, so I want to get back to

0:09:28.476 --> 0:09:31.156
<v Speaker 2>the specific sort of process narratives. So you put this

0:09:31.356 --> 0:09:35.836
<v Speaker 2>very specific prompt into GPD four, which is basically chat GPT.

0:09:36.476 --> 0:09:39.236
<v Speaker 5>I would say, it's actually better fine, and chat EPT

0:09:39.276 --> 0:09:42.156
<v Speaker 5>four is now it was better than what chatchat is

0:09:42.196 --> 0:09:43.756
<v Speaker 5>now fine for creative stuff.

0:09:44.116 --> 0:09:47.916
<v Speaker 2>Uh. Then you get some output, you get the paragraph

0:09:48.276 --> 0:09:49.596
<v Speaker 2>for it, and then what and then.

0:09:49.516 --> 0:09:52.316
<v Speaker 5>It usually it's very bad, right, And then you take

0:09:52.356 --> 0:09:54.636
<v Speaker 5>that and you put it in a program called pseudo right. Okay,

0:09:54.716 --> 0:09:57.556
<v Speaker 5>and pseudo right is a stochastic writing instrument. So you

0:09:57.556 --> 0:10:00.316
<v Speaker 5>could you then select the text and you say shorten

0:10:00.956 --> 0:10:03.716
<v Speaker 5>lengthen you say and then it has another button, which

0:10:03.756 --> 0:10:08.596
<v Speaker 5>is a customized feature, which is make it sound like X. So,

0:10:08.756 --> 0:10:11.996
<v Speaker 5>make it sound like Ernest Hemingway, make it sound like

0:10:12.196 --> 0:10:15.956
<v Speaker 5>f Scott Fitzgerald, and and and you know, the of course,

0:10:15.956 --> 0:10:17.916
<v Speaker 5>the thing I figured out very quickly is that if

0:10:17.916 --> 0:10:19.716
<v Speaker 5>you want something to sound like Margaret out With, the

0:10:19.836 --> 0:10:22.276
<v Speaker 5>very last thing you should do is put in make

0:10:22.316 --> 0:10:23.196
<v Speaker 5>it sound like markered out.

0:10:23.676 --> 0:10:24.876
<v Speaker 1>That's not enough course to me.

0:10:25.676 --> 0:10:28.476
<v Speaker 5>Well, of course, because markered Outwood is in trying to

0:10:28.476 --> 0:10:30.516
<v Speaker 5>sound like Margaret Out would she's trying to sound like

0:10:30.596 --> 0:10:34.556
<v Speaker 5>Sylvia Plathmas Philip Roth meets, it meets a bunch of

0:10:34.556 --> 0:10:34.996
<v Speaker 5>other things.

0:10:35.076 --> 0:10:37.436
<v Speaker 4>Right, then you ultimately always get back.

0:10:38.836 --> 0:10:40.356
<v Speaker 5>Yeah, And so that when you the way you get

0:10:40.396 --> 0:10:43.676
<v Speaker 5>interesting things in this text is by essentially folding these

0:10:43.756 --> 0:10:45.956
<v Speaker 5>layers of style onto each other.

0:10:46.076 --> 0:10:49.596
<v Speaker 2>Now I also use and then so pseudo right has

0:10:49.596 --> 0:10:52.116
<v Speaker 2>some output. Yeah, and then is that output what we're

0:10:52.156 --> 0:10:52.836
<v Speaker 2>reading in the book?

0:10:52.876 --> 0:10:53.196
<v Speaker 1>Correct?

0:10:54.116 --> 0:10:56.916
<v Speaker 5>Or you know, if I don't like it, I just

0:10:57.396 --> 0:11:00.396
<v Speaker 5>try again, just refresh, refresh, refresh until I guess something

0:11:00.436 --> 0:11:00.796
<v Speaker 5>that I like.

0:11:00.876 --> 0:11:03.196
<v Speaker 1>And so so this is very much a creative act.

0:11:03.196 --> 0:11:05.956
<v Speaker 4>And you're doing that basically a paragraph at a time.

0:11:06.516 --> 0:11:09.596
<v Speaker 5>Yeah, Well, with dialogue, it would go like die would

0:11:09.596 --> 0:11:12.556
<v Speaker 5>be a lot longer, right, like, because you want flow

0:11:13.196 --> 0:11:16.116
<v Speaker 5>and you want so I could do up to maybe

0:11:16.356 --> 0:11:18.636
<v Speaker 5>five hundred words of dialogue at a time. Uh huh,

0:11:18.676 --> 0:11:20.676
<v Speaker 5>So that would have been part of a much longer

0:11:20.716 --> 0:11:21.636
<v Speaker 5>series of instructions.

0:11:21.676 --> 0:11:25.156
<v Speaker 2>So this sentence human beings cannot stop making buttons, and

0:11:25.196 --> 0:11:27.356
<v Speaker 2>once we've made them, we can't stop pushing them. A

0:11:27.436 --> 0:11:30.316
<v Speaker 2>nice sentence, you know, big idea. I certainly didn't think

0:11:30.356 --> 0:11:30.796
<v Speaker 2>of that.

0:11:30.756 --> 0:11:34.076
<v Speaker 1>You didn't. It just came out of some refreshment, yeah, fresh, and.

0:11:34.156 --> 0:11:37.636
<v Speaker 5>It was in some I mean, obviously I made it,

0:11:37.676 --> 0:11:40.276
<v Speaker 5>and I authorized it too. You know, I've compared it

0:11:40.396 --> 0:11:43.356
<v Speaker 5>in the Atlantic to doing hip hop in the sense

0:11:43.396 --> 0:11:47.036
<v Speaker 5>that you're you're folding things on top of each other, right,

0:11:47.036 --> 0:11:51.836
<v Speaker 5>You're folding styles and metrics and effects on top of

0:11:51.876 --> 0:11:54.956
<v Speaker 5>each other until you get something new and weird.

0:11:55.756 --> 0:11:55.956
<v Speaker 1>Right.

0:11:56.196 --> 0:11:59.436
<v Speaker 5>And I would say about twenty times during the course

0:11:59.476 --> 0:12:02.076
<v Speaker 5>of writing it, I felt like I was, you know,

0:12:02.436 --> 0:12:05.636
<v Speaker 5>putting my hand up against something new and weird.

0:12:05.796 --> 0:12:07.396
<v Speaker 1>That's fun, right, like something.

0:12:08.676 --> 0:12:11.676
<v Speaker 5>But you know this is for most of the process,

0:12:11.956 --> 0:12:14.556
<v Speaker 5>it's just a writing tool, right, Like, it writes it

0:12:14.596 --> 0:12:17.516
<v Speaker 5>for you. You decide if it works, right, and you

0:12:17.636 --> 0:12:20.356
<v Speaker 5>tell it's you tell it what to write in.

0:12:20.316 --> 0:12:21.396
<v Speaker 4>A very granular way.

0:12:21.836 --> 0:12:24.916
<v Speaker 5>The more granular, just like writing normally, the more you

0:12:25.116 --> 0:12:27.956
<v Speaker 5>know about the bigger planning. The more planning you have

0:12:28.036 --> 0:12:30.356
<v Speaker 5>for any essay, the better the essay is going to

0:12:30.356 --> 0:12:33.436
<v Speaker 5>be right. And in this case, so you have a

0:12:33.476 --> 0:12:35.716
<v Speaker 5>plan and then you have the editing process, and in

0:12:35.836 --> 0:12:39.476
<v Speaker 5>between there's this machine. But how much of that, how

0:12:39.516 --> 0:12:43.156
<v Speaker 5>much does that matter? Is actually I don't know if

0:12:43.156 --> 0:12:45.916
<v Speaker 5>it's like twenty times it did matter where it was like,

0:12:45.956 --> 0:12:47.556
<v Speaker 5>oh that's not something I would have written, but.

0:12:47.556 --> 0:12:48.396
<v Speaker 1>It's very beautiful.

0:12:48.436 --> 0:12:51.996
<v Speaker 5>Yeah, and it's very strange, and it's you know, there's

0:12:52.036 --> 0:12:54.516
<v Speaker 5>a there's a Danish journalist who deals with go players

0:12:54.516 --> 0:12:57.916
<v Speaker 5>who play ai go against each other, and they say

0:12:57.956 --> 0:13:00.796
<v Speaker 5>it's like listening to an alien make music right, because

0:13:00.796 --> 0:13:02.716
<v Speaker 5>it's like it's not how they would play go, it's

0:13:02.756 --> 0:13:05.756
<v Speaker 5>not how a human could play go, but it's obviously

0:13:05.796 --> 0:13:09.316
<v Speaker 5>makes sense on some level. Similarly, that's how I felt

0:13:09.596 --> 0:13:12.196
<v Speaker 5>like most of the time, it's just a writing machine

0:13:12.196 --> 0:13:14.596
<v Speaker 5>that does what I tell it and then I correct it.

0:13:14.796 --> 0:13:18.436
<v Speaker 5>But then maybe twenty times you feel this new presence.

0:13:18.916 --> 0:13:19.756
<v Speaker 5>That's what's exciting.

0:13:22.476 --> 0:13:24.556
<v Speaker 2>We'll be back in a minute to hear how Lucas

0:13:24.556 --> 0:13:44.956
<v Speaker 2>Contour used AI to help him finish Schubert's unfinished symphony. Okay,

0:13:45.036 --> 0:13:48.156
<v Speaker 2>back to the conversation in Chicago with Stephen Marsh and

0:13:48.236 --> 0:13:52.916
<v Speaker 2>Lucas Contour. Lucas's story of using AI to finish Schubert's

0:13:52.996 --> 0:13:56.796
<v Speaker 2>unfinished symphony goes back to twenty nineteen. He was approached

0:13:56.796 --> 0:14:00.676
<v Speaker 2>by a Chinese tech company called Huawei. They said, we

0:14:00.756 --> 0:14:04.076
<v Speaker 2>want our phone, which runs AI, to finish Schubert's on

0:14:04.116 --> 0:14:08.076
<v Speaker 2>Finnish symphony. And they didn't know what that meant. They

0:14:08.076 --> 0:14:10.916
<v Speaker 2>had a tech team in place that was running the

0:14:10.956 --> 0:14:12.236
<v Speaker 2>AI and I knew those people.

0:14:12.236 --> 0:14:14.156
<v Speaker 6>That's why they, I think brought me in. I was

0:14:14.196 --> 0:14:16.876
<v Speaker 6>told that. So my friend, the technologist who brought me

0:14:16.916 --> 0:14:19.236
<v Speaker 6>in on this project, told me that he thought that

0:14:19.236 --> 0:14:21.156
<v Speaker 6>I would be a good fit because I have a

0:14:21.196 --> 0:14:23.916
<v Speaker 6>corporate friendly bio where they could say, oh, he can

0:14:23.996 --> 0:14:27.596
<v Speaker 6>do it. And he said, I know they you don't

0:14:27.636 --> 0:14:29.356
<v Speaker 6>have to say that part. You don't have to say

0:14:29.396 --> 0:14:31.796
<v Speaker 6>that part. He said, uh, But he said I. He

0:14:31.836 --> 0:14:34.956
<v Speaker 6>said that you, I know you can command an orchestra,

0:14:35.076 --> 0:14:37.316
<v Speaker 6>but I don't think you'll be precious about the project,

0:14:37.916 --> 0:14:39.956
<v Speaker 6>meaning that I won't be. He didn't think I would

0:14:39.956 --> 0:14:42.556
<v Speaker 6>say like, oh, well, this is heresy. We shouldn't take

0:14:42.556 --> 0:14:45.076
<v Speaker 6>Schubert's perfect work that was so perfect that he didn't

0:14:45.076 --> 0:14:50.116
<v Speaker 6>even finish it and do something with it. And uh yeah.

0:14:50.116 --> 0:14:52.996
<v Speaker 6>So I think they thought they would just that I

0:14:53.036 --> 0:14:55.356
<v Speaker 6>would press a button on the phone and a symphony

0:14:55.396 --> 0:14:58.276
<v Speaker 6>would come out and somehow a bunch of musicians would play.

0:14:58.476 --> 0:15:02.436
<v Speaker 4>So they need you for it. They just pushed the button.

0:15:02.516 --> 0:15:05.076
<v Speaker 6>So this is the conversation we had, and eventually I

0:15:05.116 --> 0:15:06.596
<v Speaker 6>had to I was on a call with them and

0:15:06.596 --> 0:15:08.356
<v Speaker 6>I said, look, this is this is not I mean,

0:15:08.356 --> 0:15:10.996
<v Speaker 6>what you're asking for in principle doesn't exist, like you

0:15:10.996 --> 0:15:13.036
<v Speaker 6>can't And I mean, what do you even want the

0:15:13.076 --> 0:15:14.676
<v Speaker 6>machine to do? Do you want it to generate audio

0:15:14.716 --> 0:15:16.276
<v Speaker 6>for you? Do you want it to generate a score?

0:15:16.516 --> 0:15:18.716
<v Speaker 6>Do you want it to perform the score? So, I mean,

0:15:18.836 --> 0:15:22.036
<v Speaker 6>right off the bat, this was a fascinating project because

0:15:22.036 --> 0:15:24.316
<v Speaker 6>I had to think about the very nature of music

0:15:24.436 --> 0:15:26.916
<v Speaker 6>to even really get started. I don't know if that

0:15:26.996 --> 0:15:29.076
<v Speaker 6>answers the question about I think it does.

0:15:29.196 --> 0:15:31.476
<v Speaker 4>I mean, I just wanted you to set yourself up,

0:15:31.476 --> 0:15:32.316
<v Speaker 4>and I think you've done it.

0:15:32.356 --> 0:15:34.436
<v Speaker 6>You want to I think I'm set up, so I'm

0:15:34.516 --> 0:15:36.436
<v Speaker 6>gonna try something new for you today. So on the

0:15:37.396 --> 0:15:41.636
<v Speaker 6>on the prep call for this event, we discussed I

0:15:41.676 --> 0:15:43.556
<v Speaker 6>said something that I don't often say out loud, but

0:15:43.556 --> 0:15:45.876
<v Speaker 6>I realized as a hallmark of my presence on stage,

0:15:45.876 --> 0:15:49.396
<v Speaker 6>is that I like to do things that might spectacularly

0:15:49.436 --> 0:15:52.636
<v Speaker 6>fail in the hopes that they will be entertaining to

0:15:52.716 --> 0:15:54.516
<v Speaker 6>an audience. So I'm going to do one of them

0:15:54.556 --> 0:15:54.796
<v Speaker 6>for you.

0:15:54.876 --> 0:15:55.036
<v Speaker 1>Now.

0:15:55.076 --> 0:15:57.116
<v Speaker 6>I'm going to I wrote a little thing about the

0:15:57.156 --> 0:16:00.476
<v Speaker 6>Unfinished Symphony. I'm going to explain it while I'm playing

0:16:00.476 --> 0:16:04.796
<v Speaker 6>some music in the background and basically scoring it as

0:16:04.836 --> 0:16:08.116
<v Speaker 6>i'm talking. So you know, wish me luck and hopefully

0:16:08.116 --> 0:16:15.556
<v Speaker 6>it'll be interesting. This is how the Unfinished Symphony starts.

0:16:29.436 --> 0:16:33.276
<v Speaker 6>A symphony has four movements, but Schubert only wrote two

0:16:33.556 --> 0:16:37.036
<v Speaker 6>and sketched a third of his eighth Symphony, the Unfinished Symphony.

0:16:38.076 --> 0:16:40.836
<v Speaker 6>No one knows why he abandoned the Unfinished Symphony, but

0:16:40.916 --> 0:16:43.916
<v Speaker 6>he did, and now it's probably his most famous work,

0:16:44.036 --> 0:16:51.236
<v Speaker 6>along with his greatest hit, Ave Maria. Some scholars believe

0:16:51.316 --> 0:16:53.716
<v Speaker 6>that Schubert couldn't find a way to fit the Eighth

0:16:53.756 --> 0:16:56.876
<v Speaker 6>Symphony into the orthodoxy of the time. Which forbade three

0:16:56.916 --> 0:16:59.596
<v Speaker 6>movements in a row in triple meter meters like three,

0:16:59.676 --> 0:17:00.436
<v Speaker 6>four and sixty eight.

0:17:01.356 --> 0:17:02.236
<v Speaker 1>But I don't believe this.

0:17:03.636 --> 0:17:06.556
<v Speaker 6>Schubert showed little reverence for orthodoxy during his short life,

0:17:07.316 --> 0:17:09.476
<v Speaker 6>and the AI that I used to finish Ubert's on

0:17:09.516 --> 0:17:21.356
<v Speaker 6>Finnish Symphony didn't believe it either. At first, we trained

0:17:21.356 --> 0:17:25.516
<v Speaker 6>the AI on recordings of Schubert's entire catalog, then prompted

0:17:25.556 --> 0:17:27.556
<v Speaker 6>it with the first two movements of the unfinished symphony.

0:17:28.036 --> 0:17:31.836
<v Speaker 6>Seems like a reasonable strategy, right, This was the result

0:17:37.876 --> 0:17:43.356
<v Speaker 6>sounds like Kat's walking on a piano, But this was

0:17:43.356 --> 0:17:48.436
<v Speaker 6>actually pretty logical. Recorded music has almost no mathematically discernible

0:17:48.436 --> 0:17:52.236
<v Speaker 6>patterns to it, so from the AI's perspective, the input

0:17:52.276 --> 0:18:04.756
<v Speaker 6>was nonsense, so more nonsense was a logical output. Music

0:18:04.756 --> 0:18:08.636
<v Speaker 6>as an abstraction is math, but music in practice is convention.

0:18:09.716 --> 0:18:13.156
<v Speaker 6>Music is understood by groups of humans, and like any language,

0:18:13.516 --> 0:18:22.876
<v Speaker 6>music doesn't have objective meaning. Music is emotionally inert left myself.

0:18:22.916 --> 0:18:26.636
<v Speaker 6>A water break is symphony. A symphony is like a skyscraper.

0:18:27.316 --> 0:18:29.956
<v Speaker 6>It's enormous, but every inch of it is designed in

0:18:29.996 --> 0:18:34.156
<v Speaker 6>meticulous detail. It's beautiful on the outside, but the inside

0:18:34.196 --> 0:18:39.436
<v Speaker 6>is filled with utilitarian solutions to simple problems. A skyscraper

0:18:40.156 --> 0:18:43.356
<v Speaker 6>has electrical columns to distribute power throughout the building, It

0:18:43.396 --> 0:18:46.316
<v Speaker 6>has plumbing, it has elevators, but you don't see any

0:18:46.316 --> 0:18:48.996
<v Speaker 6>of this essential detail when you admire the building from outside.

0:18:50.156 --> 0:18:53.876
<v Speaker 6>A symphony is like a skyscraper, but a recording of

0:18:53.916 --> 0:19:03.996
<v Speaker 6>a symphony is like a skyscraper's facade. There is no

0:19:04.076 --> 0:19:06.996
<v Speaker 6>way to tell from photos of even a million facades

0:19:07.116 --> 0:19:10.396
<v Speaker 6>that skyscrapers should have electricity, bathrooms and a way for

0:19:10.476 --> 0:19:14.516
<v Speaker 6>humans to move from one floor to another. Similarly, there

0:19:14.556 --> 0:19:16.836
<v Speaker 6>is no way to tell from the morass of frequencies

0:19:16.876 --> 0:19:19.316
<v Speaker 6>that is a piece of recorded music which frequencies are

0:19:19.356 --> 0:19:20.116
<v Speaker 6>the most important.

0:19:30.196 --> 0:19:30.676
<v Speaker 1>There we go.

0:19:33.396 --> 0:19:38.756
<v Speaker 6>So analyzing recorded music got us nowhere, and I thought

0:19:39.196 --> 0:19:42.036
<v Speaker 6>that the best way to proceed was to simplify the

0:19:42.076 --> 0:19:47.236
<v Speaker 6>task and just train the AI on the blueprints of

0:19:47.316 --> 0:19:50.316
<v Speaker 6>music rather than a finished building. So train the AI

0:19:50.356 --> 0:19:52.116
<v Speaker 6>on a blueprint rather than a finished building. So what

0:19:52.156 --> 0:19:54.236
<v Speaker 6>you just heard, what you're hearing now is the main

0:19:54.276 --> 0:19:57.236
<v Speaker 6>theme from the unfinished symphony. Here it is again, just

0:19:57.276 --> 0:20:10.876
<v Speaker 6>really listen and try to listen for the melody. And

0:20:10.956 --> 0:20:23.236
<v Speaker 6>here is that same theme reduced to its blueprint. This structure,

0:20:23.836 --> 0:20:27.796
<v Speaker 6>this blueprint in music, is just a simple melody. So

0:20:27.836 --> 0:20:29.996
<v Speaker 6>my team and I went to work extracting just the

0:20:30.036 --> 0:20:32.276
<v Speaker 6>melodies from as much of Schubert's music as we could

0:20:32.276 --> 0:20:35.716
<v Speaker 6>get our hands on. These are some examples of the

0:20:35.756 --> 0:20:42.236
<v Speaker 6>melodies we extracted. These sound robotic because they are. They

0:20:42.276 --> 0:20:45.956
<v Speaker 6>sound emotionally inert. But these are Schubert's melodies reduced to

0:20:45.996 --> 0:20:49.796
<v Speaker 6>their simplest forms, the forms that human composition students would

0:20:49.876 --> 0:20:52.876
<v Speaker 6>use when beginning a study of Schubert. Your ear knows

0:20:52.876 --> 0:20:54.836
<v Speaker 6>how to pick a melody out of a dense arrangement,

0:20:55.156 --> 0:21:02.556
<v Speaker 6>but an untrained AI cannot do this. The reason that,

0:21:02.556 --> 0:21:04.916
<v Speaker 6>since the results we wanted were simple, we needed to

0:21:04.956 --> 0:21:09.836
<v Speaker 6>train the AI on simple data. We trained on hours

0:21:09.836 --> 0:21:12.996
<v Speaker 6>of these simple melodies and then prompted again. We prompted

0:21:12.996 --> 0:21:16.196
<v Speaker 6>it with the unfinished symphony reduced to its blueprint, and

0:21:16.236 --> 0:21:23.996
<v Speaker 6>these were some of the results. So this is what

0:21:24.036 --> 0:21:27.356
<v Speaker 6>it suggested might be something that Subert would have written.

0:21:29.156 --> 0:21:31.676
<v Speaker 6>These are simple, but much more musical than the cats

0:21:31.676 --> 0:21:33.676
<v Speaker 6>walking on a piano that came from the audio only

0:21:33.716 --> 0:21:41.316
<v Speaker 6>training data. This one, for some reason, caught my attention.

0:21:41.756 --> 0:21:51.116
<v Speaker 6>Let's hear it again. I liked it, so I selected

0:21:51.116 --> 0:21:54.836
<v Speaker 6>it for embellishment. I decided to use this. I decided

0:21:54.836 --> 0:22:06.116
<v Speaker 6>to use this blueprint. This melody is a bit more

0:22:06.116 --> 0:22:09.796
<v Speaker 6>modern sounding than any of Schubert's work. If Schubert lived

0:22:09.836 --> 0:22:17.476
<v Speaker 6>to old age, these sonorities would have been available to him.

0:22:17.756 --> 0:22:21.236
<v Speaker 6>The orthodoxy around triple meters and other constraints of form

0:22:21.276 --> 0:22:25.116
<v Speaker 6>would have given way to the exploration of the Romantic period.

0:22:32.716 --> 0:22:37.276
<v Speaker 6>Providing simple singable melodies is perhaps not how most people

0:22:37.316 --> 0:22:40.716
<v Speaker 6>would imagine that an AI would be useful in writing

0:22:40.716 --> 0:22:45.236
<v Speaker 6>a symphony. But what is a symphony? Typically people think

0:22:45.236 --> 0:22:47.676
<v Speaker 6>about a symphony as something that you hear, while the

0:22:47.676 --> 0:22:53.156
<v Speaker 6>score is just a byproduct of the notated sounds. But

0:22:53.276 --> 0:22:57.956
<v Speaker 6>to me, the sound is a byproduct, and the symphony

0:22:58.036 --> 0:23:00.756
<v Speaker 6>is something that you see. It's something that you read.

0:23:01.636 --> 0:23:07.756
<v Speaker 6>It's a collection of abstract ideas in abstract notation. It's

0:23:07.876 --> 0:23:10.756
<v Speaker 6>markings on a page that serve as instructions for how

0:23:10.756 --> 0:23:16.076
<v Speaker 6>to create sounds. A symphony itself is a blueprint, and

0:23:16.156 --> 0:23:20.556
<v Speaker 6>those instructions that blueprint will be executed differently at every performance.

0:23:32.556 --> 0:23:34.236
<v Speaker 6>Let me just check out this music. It's pretty cool.

0:23:37.156 --> 0:23:40.116
<v Speaker 6>The sounds are a byproduct of the abstractions that are

0:23:40.116 --> 0:23:43.156
<v Speaker 6>expressed in the notation, and that byproduct is what the

0:23:43.196 --> 0:23:44.956
<v Speaker 6>audience experiences as a symphony.

0:23:46.116 --> 0:23:47.516
<v Speaker 1>The byproduct is what you hear.

0:23:48.796 --> 0:23:50.636
<v Speaker 6>I didn't know that I thought about music in this

0:23:50.676 --> 0:23:52.996
<v Speaker 6>way until I had to explain how I think about

0:23:53.076 --> 0:23:58.316
<v Speaker 6>music to a machine. This project taught me to question

0:23:58.396 --> 0:24:00.916
<v Speaker 6>the assumptions I make when thinking about my own craft.

0:24:04.236 --> 0:24:06.316
<v Speaker 6>I think this is the job of the AI assisted

0:24:06.356 --> 0:24:10.116
<v Speaker 6>composer today to think about what we know and to

0:24:10.156 --> 0:24:13.836
<v Speaker 6>guide our audience to rethink what happens inside their own minds.

0:24:23.276 --> 0:24:27.636
<v Speaker 6>I think it's our job to question orthodoxy. I think

0:24:27.636 --> 0:24:30.356
<v Speaker 6>it's our job to use new tools to make new art.

0:24:32.676 --> 0:24:35.756
<v Speaker 6>Today's artists are not on the verge of being replaced.

0:24:36.316 --> 0:24:39.156
<v Speaker 6>On the contrary, we are possessed of powers so great

0:24:39.276 --> 0:24:41.956
<v Speaker 6>that we will expose more truth about the human mind

0:24:42.196 --> 0:24:47.756
<v Speaker 6>and the human soul than any generation before us. We

0:24:47.836 --> 0:24:50.956
<v Speaker 6>stand on the shoulders of giants. They have given us

0:24:50.996 --> 0:24:55.716
<v Speaker 6>the language, they have given us the blueprints, they have

0:24:55.796 --> 0:24:59.556
<v Speaker 6>given us the technology. What we build with these tools

0:24:59.636 --> 0:25:02.756
<v Speaker 6>will be more powerful, and more beautiful, and more profound

0:25:02.796 --> 0:25:08.916
<v Speaker 6>than anything we can now imagine. Artificial intelligence is nothing

0:25:08.956 --> 0:25:16.156
<v Speaker 6>like us than a prosthetic for the human mind. It

0:25:16.196 --> 0:25:18.716
<v Speaker 6>will enhance art the way writing enhanced memory, the way

0:25:18.756 --> 0:25:22.396
<v Speaker 6>printing enhanced literature, the way the steam engine enhanced travel.

0:25:23.596 --> 0:25:27.836
<v Speaker 6>Artificial intelligence is an automobile. We're only beginning to emerge

0:25:27.836 --> 0:25:32.916
<v Speaker 6>from the age of horse and buggy. Artificial intelligence helped

0:25:32.956 --> 0:25:37.836
<v Speaker 6>me write the music that you're hearing right now. So

0:25:37.916 --> 0:25:39.276
<v Speaker 6>will AI kill creativity?

0:25:42.596 --> 0:25:52.436
<v Speaker 5>No, that's really rather Good's that more or less worked?

0:25:52.476 --> 0:25:53.956
<v Speaker 5>I think that's really rather good.

0:25:54.076 --> 0:25:54.396
<v Speaker 1>Thanks.

0:25:57.996 --> 0:26:00.476
<v Speaker 2>We'll be back in a minute to wave our hands

0:26:00.476 --> 0:26:14.076
<v Speaker 2>a little bit about the future of AI and creativity.

0:26:14.276 --> 0:26:15.276
<v Speaker 6>That's the end of the ads.

0:26:15.716 --> 0:26:16.836
<v Speaker 1>Now we're going back to the show.

0:26:17.716 --> 0:26:19.756
<v Speaker 5>The reason I knew AI was going to take off

0:26:20.316 --> 0:26:21.956
<v Speaker 5>was when I was writing a piece for The New

0:26:21.996 --> 0:26:26.236
<v Speaker 5>Yorker about GPT three and I got it to finish

0:26:26.396 --> 0:26:33.436
<v Speaker 5>off Coleridge's Kubla Khan is great unfinished poem, and it

0:26:33.516 --> 0:26:36.716
<v Speaker 5>did it perfectly well. Like I mean, if somebody told me, yeah,

0:26:36.716 --> 0:26:38.996
<v Speaker 5>this is how it ended, I would have been like, great,

0:26:39.396 --> 0:26:41.596
<v Speaker 5>right and so, And it did it like that like

0:26:41.756 --> 0:26:42.276
<v Speaker 5>one second.

0:26:42.316 --> 0:26:43.716
<v Speaker 1>I mean, it was just so incredible to me.

0:26:43.996 --> 0:26:46.556
<v Speaker 2>Just to sort of close this part of the conversation,

0:26:47.076 --> 0:26:50.436
<v Speaker 2>I'm curious. I mean, both of these projects. We were

0:26:50.596 --> 0:26:54.316
<v Speaker 2>very AI forward, right, They were like high concept, you know,

0:26:54.476 --> 0:26:57.396
<v Speaker 2>sort of let's explicitly wrap this thing in AI.

0:26:57.916 --> 0:26:58.236
<v Speaker 4>Fine.

0:26:58.436 --> 0:27:02.956
<v Speaker 2>Interesting, But presumably the real action comes in the things

0:27:02.956 --> 0:27:05.516
<v Speaker 2>that are just what you guys are working on that

0:27:05.676 --> 0:27:07.796
<v Speaker 2>just happens to have AI as a tool, the same

0:27:07.836 --> 0:27:09.836
<v Speaker 2>way say a Google search, which by the way, is

0:27:09.876 --> 0:27:12.516
<v Speaker 2>a kind of AI, is also a tool, right, And

0:27:12.596 --> 0:27:16.316
<v Speaker 2>so I'm curious in your work now on other projects

0:27:16.676 --> 0:27:18.756
<v Speaker 2>that are not like, hey, look this was made with

0:27:18.796 --> 0:27:21.356
<v Speaker 2>AI kind of projects. Are you guys using AI? And

0:27:21.356 --> 0:27:23.996
<v Speaker 2>if so, how what do you want to go first?

0:27:24.076 --> 0:27:27.916
<v Speaker 6>Yeah? Yeah, first, so yeah, obviously of course, like it's

0:27:27.916 --> 0:27:29.796
<v Speaker 6>in everybody's pockets, you use it all the time. And

0:27:30.636 --> 0:27:35.036
<v Speaker 6>AI has done nothing so far other than help my career.

0:27:35.036 --> 0:27:37.636
<v Speaker 6>And I don't mean just by doing this, which was fantastic.

0:27:37.676 --> 0:27:39.556
<v Speaker 6>But when I write a piece of music and put

0:27:39.596 --> 0:27:42.436
<v Speaker 6>it on Spotify, the reason you hear it is because

0:27:42.436 --> 0:27:44.596
<v Speaker 6>an AI recommended it to you. You know, that's the

0:27:44.636 --> 0:27:46.556
<v Speaker 6>only reason you're going to find it. And so and

0:27:46.996 --> 0:27:49.556
<v Speaker 6>these types of algorithms that are generating that are keeping

0:27:49.556 --> 0:27:52.556
<v Speaker 6>people out on apps longer and keeping people on Netflix

0:27:52.796 --> 0:27:56.036
<v Speaker 6>and on Spotify longer, are putting money not enough money,

0:27:56.076 --> 0:27:59.796
<v Speaker 6>and that's another panel discussion, but putting money in our

0:27:59.836 --> 0:28:00.716
<v Speaker 6>pockets directly?

0:28:00.836 --> 0:28:03.356
<v Speaker 2>Let me let me ask a more precise version of

0:28:03.396 --> 0:28:05.836
<v Speaker 2>the question in response to that clever answer.

0:28:05.996 --> 0:28:07.356
<v Speaker 4>Do you use generative AI?

0:28:08.716 --> 0:28:08.956
<v Speaker 1>Yes?

0:28:09.596 --> 0:28:12.556
<v Speaker 6>And also this is a terminology problem.

0:28:12.476 --> 0:28:15.396
<v Speaker 1>But you know what do you use music?

0:28:15.436 --> 0:28:19.076
<v Speaker 2>Do you use AI to generate musical ideas for you?

0:28:19.236 --> 0:28:19.476
<v Speaker 1>Yes?

0:28:19.796 --> 0:28:22.076
<v Speaker 6>But also like what is a musical idea? I use

0:28:22.116 --> 0:28:24.356
<v Speaker 6>a parametric eque that I mean they were using a

0:28:24.516 --> 0:28:28.236
<v Speaker 6>they were using this was there was probably good. I'm trying,

0:28:28.276 --> 0:28:30.276
<v Speaker 6>well the answer the answer is yes.

0:28:30.316 --> 0:28:32.236
<v Speaker 4>I know what you're saying. But I feel like you

0:28:32.316 --> 0:28:33.076
<v Speaker 4>know what I'm saying.

0:28:33.236 --> 0:28:35.476
<v Speaker 6>Well, yes, I'm The reason I'm trying to drill down

0:28:35.516 --> 0:28:38.156
<v Speaker 6>here is because this there tell me how to ask

0:28:38.196 --> 0:28:41.516
<v Speaker 6>the question I want to asking doesn't have the answer

0:28:41.596 --> 0:28:42.036
<v Speaker 6>that you want?

0:28:42.996 --> 0:28:47.036
<v Speaker 2>Right, So fair, what's the what? What's the smarter version

0:28:47.076 --> 0:28:49.196
<v Speaker 2>of the question? I'm not well enough equipped to ask.

0:28:50.756 --> 0:28:51.556
<v Speaker 1>I don't know if I can.

0:28:51.676 --> 0:28:52.676
<v Speaker 6>I don't know if I can help you with that.

0:28:53.156 --> 0:28:58.596
<v Speaker 1>I don't let.

0:28:58.516 --> 0:28:59.596
<v Speaker 4>Me ask the question to you.

0:29:00.596 --> 0:29:02.036
<v Speaker 1>Thank you for your Stephen.

0:29:02.716 --> 0:29:05.636
<v Speaker 2>Do you you use generative AI when you're writing with

0:29:05.836 --> 0:29:06.276
<v Speaker 2>other things?

0:29:06.276 --> 0:29:08.236
<v Speaker 5>Okay, here's the thing, and I think this is sort

0:29:08.236 --> 0:29:10.676
<v Speaker 5>of where we're going. Like I would when I write

0:29:10.756 --> 0:29:13.356
<v Speaker 5>something for a magazine or newspaper or novel that I'm

0:29:13.356 --> 0:29:15.716
<v Speaker 5>working on, I would never use chatchipt.

0:29:15.316 --> 0:29:18.236
<v Speaker 1>Even to get an idea because here or whatever they

0:29:18.916 --> 0:29:22.076
<v Speaker 1>because I'm so much smarter than chat GPT.

0:29:22.276 --> 0:29:24.876
<v Speaker 5>Right, And I'm like when you and what you have

0:29:24.916 --> 0:29:27.036
<v Speaker 5>to also have to understand is chatchypt. The reason it's

0:29:27.036 --> 0:29:30.436
<v Speaker 5>so successful is exactly that it has been banalified, like

0:29:30.476 --> 0:29:33.596
<v Speaker 5>when you use other generative ais that we have access to,

0:29:33.716 --> 0:29:37.556
<v Speaker 5>because you realize that like these are the ones that

0:29:37.596 --> 0:29:41.476
<v Speaker 5>the public uses are very poor creatively, like they're actually.

0:29:41.156 --> 0:29:43.276
<v Speaker 4>But you have access to the good ones, to the

0:29:43.276 --> 0:29:43.756
<v Speaker 4>good stuff.

0:29:43.756 --> 0:29:45.716
<v Speaker 1>Here's the thing you can't get on when when you

0:29:45.836 --> 0:29:47.036
<v Speaker 1>use the good stuff.

0:29:46.836 --> 0:29:49.196
<v Speaker 5>What the good stuff is going to be used for

0:29:49.316 --> 0:29:52.876
<v Speaker 5>stuff that doesn't exist yet. What we're seeing here is

0:29:52.916 --> 0:29:56.476
<v Speaker 5>the birth of a new medium, right and what and

0:29:56.556 --> 0:30:00.796
<v Speaker 5>so when it comes to write an essay, what people

0:30:00.836 --> 0:30:03.156
<v Speaker 5>want when they write, when they read an essay, is

0:30:03.156 --> 0:30:07.116
<v Speaker 5>a human being communicating their thoughts and feelings, right, they

0:30:07.116 --> 0:30:09.756
<v Speaker 5>don't want like they don't That's why they go to it.

0:30:09.916 --> 0:30:13.876
<v Speaker 5>And a generative AI cannot do that generative Like it's

0:30:13.876 --> 0:30:16.116
<v Speaker 5>sort of like asking, like do you use film to

0:30:16.196 --> 0:30:19.556
<v Speaker 5>make theater? Like at first, you know, when you when

0:30:19.636 --> 0:30:22.116
<v Speaker 5>film was invented, all they did was cannibalized theater and

0:30:22.116 --> 0:30:24.596
<v Speaker 5>they were putting on weird shows or they were recreating

0:30:24.636 --> 0:30:27.436
<v Speaker 5>news events and things like this. That's where we're at

0:30:27.476 --> 0:30:29.636
<v Speaker 5>right now. This is going to be used for new

0:30:29.836 --> 0:30:34.316
<v Speaker 5>art forms that don't exist, and that's that's the exciting stuff.

0:30:34.356 --> 0:30:36.436
<v Speaker 5>And it's also why it's almost impossible to do.

0:30:36.676 --> 0:30:38.596
<v Speaker 2>You mean, like the book that is never done, the

0:30:38.596 --> 0:30:40.516
<v Speaker 2>book where it can or like what like.

0:30:40.476 --> 0:30:42.556
<v Speaker 5>I'm written that I have written a short story that

0:30:42.716 --> 0:30:45.036
<v Speaker 5>is infinite art forms?

0:30:45.076 --> 0:30:46.636
<v Speaker 1>Like what do you have in your mind when you

0:30:46.676 --> 0:30:46.996
<v Speaker 1>say it?

0:30:47.276 --> 0:30:50.636
<v Speaker 5>Well, like, for example, I'm working with cohere to recreate

0:30:51.276 --> 0:30:54.156
<v Speaker 5>the Oracle at Delphi. Right there's a large amount of

0:30:54.156 --> 0:30:56.276
<v Speaker 5>information that you can glean from that, and there's also

0:30:56.356 --> 0:30:58.076
<v Speaker 5>pretty interesting historical record.

0:30:58.796 --> 0:31:01.716
<v Speaker 4>And so you'll ask it a question and it will answer, yes.

0:31:01.636 --> 0:31:04.236
<v Speaker 5>We're try and recreate the experience of going to the

0:31:04.236 --> 0:31:06.876
<v Speaker 5>Oracle at Delphia as closely as we can use effects.

0:31:07.036 --> 0:31:09.596
<v Speaker 6>Yeah, it's a perfect use of AI and so oracles.

0:31:09.756 --> 0:31:11.116
<v Speaker 6>This is one of the things that has come up

0:31:11.116 --> 0:31:13.596
<v Speaker 6>in my research is that we use oracles because we're

0:31:13.636 --> 0:31:16.636
<v Speaker 6>bad at doing things randomly. So if we're out in

0:31:16.676 --> 0:31:19.316
<v Speaker 6>the wilderness, we'll just go hunt in the same place

0:31:19.396 --> 0:31:22.276
<v Speaker 6>over and over and over again, right, And eventually animals

0:31:22.276 --> 0:31:24.196
<v Speaker 6>figure it out and they say, just don't hang out there.

0:31:23.996 --> 0:31:25.636
<v Speaker 1>And you won't get eaten by the humans.

0:31:25.676 --> 0:31:28.036
<v Speaker 6>And so when we like consult an oracle, or roll

0:31:28.116 --> 0:31:30.356
<v Speaker 6>some dice, or like ask the sacred chickens if we

0:31:30.356 --> 0:31:32.916
<v Speaker 6>should go to war, they're basically giving us a random answer.

0:31:33.036 --> 0:31:36.116
<v Speaker 1>That's right. There are randomization engines, see, and it's.

0:31:35.996 --> 0:31:38.756
<v Speaker 5>Things of this nature that I think will be that

0:31:39.236 --> 0:31:42.316
<v Speaker 5>I'm excited about to use it. We're cannibalizing forms. That's

0:31:42.356 --> 0:31:44.516
<v Speaker 5>what I do writing short stories to It's very interesting.

0:31:44.796 --> 0:31:46.716
<v Speaker 5>But the truth is that what this can be used

0:31:46.796 --> 0:31:48.956
<v Speaker 5>for we don't know yet, and what it's going to

0:31:48.996 --> 0:31:51.876
<v Speaker 5>be used for is some weird and the problem is

0:31:51.916 --> 0:31:55.556
<v Speaker 5>there's absolutely no institutions to do it with, right, Like.

0:31:55.956 --> 0:31:59.076
<v Speaker 4>Nobody will buy your oracle of Veli's.

0:31:58.556 --> 0:32:00.396
<v Speaker 1>Supposed to take oracle of Hi.

0:32:00.876 --> 0:32:03.356
<v Speaker 5>My name Stephen, I'd like to recreate the oracle at

0:32:03.356 --> 0:32:07.716
<v Speaker 5>DELFI using generative AI. I'm sorry, sir, this is a

0:32:07.876 --> 0:32:11.396
<v Speaker 5>key mark, you know what I mean, like like it

0:32:11.636 --> 0:32:14.116
<v Speaker 5>like it's not that's that's not like there's no one

0:32:14.196 --> 0:32:18.156
<v Speaker 5>to go to. So that's that's where we're at. To me, Like,

0:32:18.236 --> 0:32:20.116
<v Speaker 5>I think the the the thing that I think is

0:32:20.196 --> 0:32:23.596
<v Speaker 5>very obvious is that when you use generative AI, what

0:32:23.676 --> 0:32:27.996
<v Speaker 5>it is very good at is the most stock answer, right.

0:32:28.156 --> 0:32:30.836
<v Speaker 5>And that's why it's so such a threat to like

0:32:30.956 --> 0:32:34.356
<v Speaker 5>the undergraduate essay, right, because that there you're basically looking

0:32:34.396 --> 0:32:38.956
<v Speaker 5>for the fulfillment of a stylistic you know, set pattern

0:32:39.436 --> 0:32:40.236
<v Speaker 5>that it can do.

0:32:40.716 --> 0:32:40.996
<v Speaker 1>Right.

0:32:41.036 --> 0:32:46.116
<v Speaker 5>But people respond to human like there's this weird idea

0:32:46.156 --> 0:32:49.156
<v Speaker 5>that art is something external to our experience of it.

0:32:49.156 --> 0:32:52.636
<v Speaker 5>It isn't. It's just we we have we create tools.

0:32:52.676 --> 0:32:54.956
<v Speaker 5>As the moment we find tools, all we're thinking of

0:32:55.156 --> 0:32:57.436
<v Speaker 5>is can we do something weird with it? And I think,

0:32:57.636 --> 0:32:59.516
<v Speaker 5>I mean, one thing that I've really learned doing this

0:32:59.676 --> 0:33:05.036
<v Speaker 5>is that creativity is instructible. Like it it doesn't matter

0:33:05.076 --> 0:33:09.356
<v Speaker 5>what comes down technologically, what comes down politically, what Like,

0:33:10.036 --> 0:33:13.516
<v Speaker 5>we are creative animals and we have to understand that

0:33:13.516 --> 0:33:17.396
<v Speaker 5>that's just our nature and nothing is gonna kill it, nothing,

0:33:17.596 --> 0:33:19.196
<v Speaker 5>not certainly not chat gept.

0:33:19.596 --> 0:33:21.996
<v Speaker 6>Great I can I can sum up the history of

0:33:22.076 --> 0:33:25.956
<v Speaker 6>music from the year sixty thousand before present to now

0:33:26.316 --> 0:33:28.196
<v Speaker 6>with one sentence, and maybe you'll agree that this sums

0:33:28.236 --> 0:33:30.196
<v Speaker 6>up the history of art already. It's the search for

0:33:30.276 --> 0:33:33.236
<v Speaker 6>new sounds. Yeah, that's it. That's all there is to it.

0:33:33.276 --> 0:33:36.516
<v Speaker 6>If something exists, nobody cares and chat geept I will chat.

0:33:36.596 --> 0:33:39.476
<v Speaker 6>Chapet doesn't do music. But there are many music generative ais,

0:33:39.836 --> 0:33:44.196
<v Speaker 6>and they generate music that, like charitably would call insipid.

0:33:44.556 --> 0:33:47.396
<v Speaker 6>Yeah it's fine, like it's music. You would recognize it

0:33:47.436 --> 0:33:49.596
<v Speaker 6>as music, but nobody. You wouldn't listen to it. It'll

0:33:49.636 --> 0:33:53.636
<v Speaker 6>get bad music. It won't, so it'll sound better, it'll

0:33:53.676 --> 0:33:56.276
<v Speaker 6>sound better. So this is the but, but nobody cares

0:33:56.276 --> 0:33:58.316
<v Speaker 6>about that. So as soon as like, as soon as

0:33:58.356 --> 0:34:01.316
<v Speaker 6>you can have so Jacob for your podcast, as soon

0:34:01.316 --> 0:34:04.316
<v Speaker 6>as you can have beautiful sounding orchestral music like this

0:34:04.636 --> 0:34:07.396
<v Speaker 6>for free, you're gonna want something else because this is

0:34:07.436 --> 0:34:09.796
<v Speaker 6>available and it's everywhere, and so what you're gonna what

0:34:09.876 --> 0:34:12.116
<v Speaker 6>you're gonna want is like the thing where like Lucas

0:34:12.156 --> 0:34:14.716
<v Speaker 6>plays a guitar with a really nice sounding reverb. That's

0:34:14.756 --> 0:34:17.516
<v Speaker 6>gonna be the style and you can trace and we

0:34:17.556 --> 0:34:19.156
<v Speaker 6>have a we have a composer in the audience who could,

0:34:19.156 --> 0:34:21.076
<v Speaker 6>hopefully will agree with me on this, and a professor

0:34:21.116 --> 0:34:23.356
<v Speaker 6>of this kind of thing. But you can trace musical

0:34:23.356 --> 0:34:26.236
<v Speaker 6>styles in media, and it's like whatever is ubiquitous just

0:34:26.276 --> 0:34:28.516
<v Speaker 6>falls out of fashion and then that whatever the opposite

0:34:28.556 --> 0:34:33.316
<v Speaker 6>of it is becomes becomes fashionable. So yeah, that's my

0:34:33.476 --> 0:34:35.516
<v Speaker 6>that's my two cents the search for new sounds.

0:34:36.796 --> 0:34:48.196
<v Speaker 1>Thanks you guys. This is closure. Yeah.

0:34:48.636 --> 0:34:53.196
<v Speaker 2>My conversation with Lucas Contour and Stephen Marsh was organized.

0:34:52.756 --> 0:34:54.316
<v Speaker 1>By Chicago Humanities.

0:34:56.756 --> 0:35:00.916
<v Speaker 2>Today's show was edited by Karen Chakerji, produced by Edith Russolo,

0:35:01.076 --> 0:35:03.316
<v Speaker 2>and engineered by Amanda k Wong.

0:35:04.556 --> 0:35:05.436
<v Speaker 1>You can email us.

0:35:05.356 --> 0:35:08.916
<v Speaker 2>At Problem at pushkin dot fm. We are always, always,

0:35:09.116 --> 0:35:12.116
<v Speaker 2>always trying to find interesting new guests for the show,

0:35:12.196 --> 0:35:14.156
<v Speaker 2>So if there's somebody who think we should book, please

0:35:14.236 --> 0:35:16.876
<v Speaker 2>let us know. I'm Jacob Goldstein and we'll be back

0:35:16.956 --> 0:35:34.356
<v Speaker 2>next week with another episode of What's Your Problem.