AI transcript
0:00:02 (upbeat music)
0:00:06 On June 27th, our team headed to New York City.
0:00:08 We are at the A16Z office
0:00:11 for the first ever AI artist retreat.
0:00:15 That was A16Z consumer partner, Justine Moore.
0:00:16 Justine was one of many partners
0:00:18 who attended this retreat,
0:00:19 which brought together the builders
0:00:22 behind some of the most popular AI creative tools
0:00:23 in existence.
0:00:28 That is, 11 labs, Korea, FIGL, Udio, ideogram, and civet.
0:00:31 All together with 15 top artists.
0:00:34 These are the folks who are often
0:00:36 doing the coolest things with these sorts of tools.
0:00:38 They’re kind of pushing the boundaries
0:00:40 of what the tools can create.
0:00:43 Today, you’ll get to hear from many of these AI founders,
0:00:45 who together with these artists
0:00:48 are advancing what it means to be creative.
0:00:50 Art is going to get better than ever.
0:00:53 The average art output is going to improve,
0:00:55 but so is the ceiling.
0:00:57 It also is a higher participation rate.
0:00:59 Everyone who’s interested in creativity
0:01:01 can be creative and express themselves,
0:01:02 which is just so cool.
0:01:04 That was Anish Acharya,
0:01:06 general partner on the consumer team,
0:01:08 but that’s not all.
0:01:09 I’ve been a founder twice.
0:01:12 I’ve been spinning records as a DJ for 25 years,
0:01:14 and I’m all about AI and art.
0:01:17 So what happens when you put all these investors,
0:01:20 leading artists and creative tool founders,
0:01:21 all into the same room?
0:01:24 I mean, the vibes have been immaculate.
0:01:25 And I think that the thing that’s the most surprising
0:01:27 is how much everyone has in common.
0:01:29 Like the founders are more creative,
0:01:33 and the creatives and artists are more technical.
0:01:33 I think the other thing
0:01:36 has just been how interdisciplinary it all is.
0:01:37 People making video,
0:01:39 want to play with generative audio,
0:01:40 people making music,
0:01:42 want to play with sound effects.
0:01:44 It’s just incredible to see.
0:01:45 One of the coolest things was
0:01:48 a lot of the founders had recognized people
0:01:50 by their online screen names or knew,
0:01:52 “Oh my gosh, you used my tool
0:01:55 to create this incredible song that went super viral.”
0:01:57 Or, “You used my product to make this
0:01:59 kind of amazing video animation
0:02:01 that our whole team was talking about for a week.”
0:02:03 These are people who have been interacting with each other
0:02:08 often daily online for the past six, 12, 18 months,
0:02:09 sometimes even two years,
0:02:13 but didn’t even know what each other looked like in person.
0:02:16 Now today, you get a behind the scenes look into this event,
0:02:19 including the origin stories behind many of these tools,
0:02:22 which by the way, some have never been shared publicly,
0:02:24 and have these tools,
0:02:26 which have all gone through their own viral moments,
0:02:28 are navigating this AI wave,
0:02:30 and what they see on the horizon.
0:02:31 Let’s get started.
0:02:35 As a reminder,
0:02:38 the content here is for informational purposes only,
0:02:40 should not be taken as legal, business, tax,
0:02:41 or investment advice,
0:02:43 or be used to evaluate any investment or security,
0:02:46 and is not directed at any investors or potential investors
0:02:48 in any A16Z fund.
0:02:50 Please note that A16Z and its affiliates
0:02:51 may also maintain investments
0:02:54 in the companies discussed in this podcast.
0:02:56 For more details, including a link to our investments,
0:02:59 please see a16z.com/disclosures.
0:03:06 Here we are in 2024.
0:03:08 We’re at an exciting inflection
0:03:10 where your creativity is being unbounded
0:03:12 by the tools available.
0:03:15 – I mean, we’re early, but there’s more people
0:03:17 making more art and more people making more tools
0:03:20 to make art than ever before.
0:03:23 And if you kinda look at the history of technology and art,
0:03:25 every single time there’s been a new technology,
0:03:28 the amount of art has dramatically increased.
0:03:31 People worry that drum machines would compete with drummers,
0:03:33 and instead there’s more people making more music
0:03:35 with both drummers and drum machines than ever before,
0:03:37 so I think there’s a sort of equivalent moment
0:03:39 here in technology and art,
0:03:42 where we’re at the beginning of everybody
0:03:45 who has taste and interest in art being able to make it.
0:03:48 – Many have drawn parallels to prior computing waves,
0:03:50 but is this any different?
0:03:51 – Well, what’s different is for the first time
0:03:53 we’re creating these sort of left brain things.
0:03:54 You know what I mean?
0:03:56 Computers and computing platforms
0:03:58 have really been in the business of precision,
0:04:00 and now we’re creating products
0:04:03 that are intentionally imprecise, beautifully imprecise,
0:04:06 so it just feels like a whole different flavor
0:04:09 for products and product design than we’ve ever seen before.
0:04:11 – So let’s introduce you to some of the people
0:04:12 behind those products.
0:04:14 – We have companies here covering basically
0:04:18 every sort of creative modality, image, video, music,
0:04:20 3D speech, all those sorts of things.
0:04:21 – That includes–
0:04:23 – Connor, I’m a co-founder at UDO.
0:04:24 – And–
0:04:26 – Omar, I’m the head of design at Love & Labs.
0:04:28 – Both companies are focused on audio,
0:04:30 with UDO focused on music,
0:04:32 while Love & Labs is tackling everything
0:04:34 from voice to sound effects.
0:04:35 Meanwhile, founders like–
0:04:39 – Mohammed, I’m the co-founder CEO at Ideogram.
0:04:40 – Victor and–
0:04:41 – Diego.
0:04:42 – Who are the co-founders of–
0:04:42 – Greya.
0:04:43 – And–
0:04:45 – Hong, I’m working on VEGLE.
0:04:46 – These founders are building
0:04:48 at the increasingly sophisticated world
0:04:51 of 2D imagery and video, plus 3D.
0:04:54 Ideogram, for example, lets you generate AI imagery
0:04:56 with accurate text embedded,
0:04:59 a surprisingly difficult technical feat.
0:05:00 VEGLE, on the other hand, is building
0:05:03 at the intersection of video and 3D.
0:05:06 Meanwhile, Greya’s come up with a suite of AI tools,
0:05:08 like upscalers and real-time generation.
0:05:12 Or, in the case of Civit, a new breed of marketplace.
0:05:14 – My name’s Maxwell Holker.
0:05:17 I am COO at Civiti and co-founder.
0:05:18 – Yes, and I’m Justin Mayer.
0:05:21 I’m the COO and co-founder of Civit as well.
0:05:22 – And CPO and CTO and–
0:05:24 – Lots of things, the joys of a startup.
0:05:26 We are a massive community of people
0:05:29 making tons and tons of AI creations,
0:05:33 using community-made models with community-made patches
0:05:35 to those models called LORAS.
0:05:37 We give people the ability to either train
0:05:40 on a few specific models, so a model focused on anime,
0:05:42 or a model focused on being semi-realistic.
0:05:44 Or they can select their own custom model
0:05:46 to train on top of.
0:05:48 – With AI moving so quickly, it’s clear
0:05:50 that we no longer live in a world
0:05:52 of just chat GBT and mid-journey.
0:05:55 Numerous companies have springboarded into the zeitgeist
0:05:57 and grown at unprecedented rates.
0:06:00 So we thought it was fitting to take a step back
0:06:02 and document this whirlwind of a journey.
0:06:04 While many of these founders have been quietly working
0:06:08 in research for years, their origin story often started
0:06:10 from scratching their own itch.
0:06:11 Muhammad from I8Gram.
0:06:15 – I guess part of it is that there is this thesis
0:06:19 that everybody has an innate desire to create.
0:06:24 And as humans, we have this inner creative child.
0:06:28 The education system sometimes kills
0:06:30 this creative child, unfortunately.
0:06:34 And what’s finally possible with technology and AI
0:06:38 is to help people express themselves visually
0:06:39 and creatively.
0:06:40 So that’s the interesting part.
0:06:45 When you think of using image for communication,
0:06:48 then you can communicate much more effectively
0:06:51 if you have image and text together.
0:06:54 – For Muhammad, it really was this unique combination
0:06:56 of text and imagery.
0:06:59 – For me, image and video is dear to my heart
0:07:01 and very personal.
0:07:03 – But for Connor, it was his connection to music.
0:07:06 – Music for me, I think, is very special medium.
0:07:07 It’s everywhere at all times.
0:07:09 Like it’s in the background when you’re at a restaurant
0:07:11 or a cafe, you’re listening to your headphones
0:07:12 and you’re going to work in the morning.
0:07:15 It really has an emotional resonance with people.
0:07:17 And for me, making that abundance,
0:07:19 like the kind of promise of generative modeling
0:07:21 is that a lot of this can be far more abundant
0:07:23 than it ever was before.
0:07:24 – And for Victor, it was his discovery
0:07:27 that programming itself was the creative gateway.
0:07:29 – When I discovered about programming,
0:07:32 that was great to me because I realized that through coding,
0:07:33 you can also be super creative.
0:07:37 But the moment where I discovered about early gen AI models
0:07:40 like DCGan and later on, StyleGan,
0:07:41 that’s when my mind was blown.
0:07:45 And when I realized about the creative potential
0:07:46 that this technology had,
0:07:48 and that’s when I fell into the rabbit hole.
0:07:49 And I feel like Korea, to me,
0:07:52 it’s been kind of the snowball that it has started
0:07:55 with me realizing that you can use artificial intelligence
0:07:57 in a creative way.
0:07:59 – But for Omar, it was building his own side projects
0:08:01 and a desire to share what he was learning
0:08:04 that actually propelled him into his role at 11 Labs.
0:08:07 – It’s really funny, actually.
0:08:10 I, over the last maybe couple of years,
0:08:13 started diving into AI tools when ChatGBT came out
0:08:15 and started making things on the side for fun.
0:08:17 One of those was a children’s book
0:08:19 that ended up accidentally going viral.
0:08:22 And that kind of was my journey into AI.
0:08:23 Through that and making that book,
0:08:25 I started exploring other AI tools.
0:08:29 And what I really enjoyed was sharing what I was doing
0:08:30 and how I made it.
0:08:33 So I discovered 11 Labs and made a podcast with 11 Labs
0:08:36 where I was talking to a fictional figure
0:08:38 and we were having a back and forth conversation
0:08:40 that also kind of did the numbers on Twitter.
0:08:42 And then I was like, I love using this tool.
0:08:45 I’m gonna make my own AI short movie.
0:08:46 Actually, Justine and I are friends.
0:08:48 And so I showed it to her and I was like,
0:08:51 I kind of need free credits because this movie
0:08:53 is using up all the credits on 11 Labs.
0:08:55 And she’s like, you should meet the founder, Maddie.
0:08:57 We met, we really hit it off.
0:08:59 And Maddie, in classic Maddie fashion,
0:09:02 was very direct and at the end of the call was like,
0:09:05 hey, we’re actually looking to hire someone to lead design.
0:09:06 Are you interested?
0:09:09 And then to work on a product that I’d used for over a year.
0:09:10 That experience also gave Amar
0:09:13 a taste of just how quickly this space moves
0:09:15 and also a hit of virality.
0:09:18 It happened because a friend of mine had their first kid
0:09:21 and I read her children’s book actually.
0:09:21 I was reading it and I was like,
0:09:23 this story kind of makes no sense.
0:09:25 So, so I went back home.
0:09:26 I’d been using the journey a lot,
0:09:28 chat to you two weeks old,
0:09:30 combine the two to create that book.
0:09:33 And then I was like, how do I get this published?
0:09:35 And Amazon has this amazing publishing service.
0:09:37 You can get a book out within 48 hours.
0:09:41 I had a paper back in my hand in 72 hours, so fast.
0:09:44 And it’s really interesting because writing a book
0:09:46 and publishing on Amazon is like,
0:09:48 it was almost like iterating on software.
0:09:49 If I discovered a type or whatever,
0:09:51 I just updated the PDF and the new book was out
0:09:53 and a new publishing line was out.
0:09:56 And so, yeah, I put it out there,
0:09:57 got a ton of virality from that.
0:10:01 And yeah, that was a really interesting experience.
0:10:04 – Pre-AI, we were in this era of consumer
0:10:07 where it was just really hard to get people’s attention,
0:10:09 really hard to get them to download a new app
0:10:10 or try a new tool.
0:10:13 You had to spend a lot of money on customer acquisition.
0:10:16 Now, just with the real excitement around AI,
0:10:18 if you make a cool product,
0:10:20 you can get it into the hands of people
0:10:22 and get them using and talking about it.
0:10:25 – This was the case for Victor and Diego at Cria,
0:10:28 who eventually met their own viral moment,
0:10:29 although it didn’t come easy.
0:10:32 – First of all, it was called Geniverse,
0:10:36 coming from Generative Universe, best name ever.
0:10:38 And essentially, it was like two things.
0:10:40 It was on the one side, an open source library
0:10:43 that it was kind of integrating all the cool stuff
0:10:45 that it was available at that moment.
0:10:47 And on the other side, it was a creative tool.
0:10:49 And the way how it looked, it was like super experimental.
0:10:52 Like, we didn’t really know how to do UI design
0:10:53 or any of that.
0:10:56 Like, the background really had stars and everything.
0:10:59 So it was like a galaxy, like the Generative Universe, right?
0:11:03 And then you could put text, you could put images,
0:11:05 and you had a few things that you could tweak
0:11:06 and you could generate images.
0:11:09 And you would see like the image evolving in real time.
0:11:12 And the images that you liked, you could keep them.
0:11:14 And they were added to this kind of universe.
0:11:17 And essentially, you ended up with a ton of images
0:11:18 in this interactive space.
0:11:22 So for us, it was always with the same idea in mind,
0:11:24 on the one side, controllability.
0:11:26 And on the other side, intuitiveness.
0:11:30 Like, how do we make tools that doesn’t look daunting?
0:11:32 Because AI, in the end, is like a new creative medium.
0:11:34 A lot of people are using it for the first time.
0:11:37 And we want them to have the inexperience
0:11:39 where the AI does what you expect to do.
0:11:42 And you don’t need to learn about crazy prom and generings
0:11:46 and all these tweaks up to get good results.
0:11:47 And on the other side is controllability
0:11:49 because we are dealing with creatives.
0:11:52 We are dealing with folks who are not just OK
0:11:54 with having a beautiful image.
0:11:55 They want that beautiful image.
0:11:59 So these are the two core principles that we had since then.
0:12:03 And we built kind of a Figma-ish interface for AI.
0:12:07 And we had every single utility that you had at that point
0:12:08 with the stable diffusion in there.
0:12:11 We had like thousands of AI models that you could use.
0:12:13 We had every single technique, like every control net.
0:12:15 Everything was in there.
0:12:16 But you know, it was not working.
0:12:18 It was like a learning curve that some people
0:12:20 were just not willing to take.
0:12:24 So then we have the first kind of virality moment
0:12:27 when we ship this thing that it was almost like an equivalent
0:12:28 to a meme generator.
0:12:31 I remember that we were seeing all of these images on Twitter
0:12:33 with the spirals, right?
0:12:35 So we were like, well, what’s going on with these spirals?
0:12:36 Like, we can do it.
0:12:38 This is like one day of work.
0:12:40 And I remember at that point, Diego was like,
0:12:41 we should do something with this.
0:12:43 We should do something with this.
0:12:44 It’s getting so viral.
0:12:45 And I was more like in the mood of, like,
0:12:48 we need to ship this whatever feature we were working on.
0:12:50 At that moment, until at one point, we were like,
0:12:51 OK, let’s fucking do it.
0:12:53 And we did it like in the sketchiest way possible,
0:12:55 like in one or two days.
0:12:57 And we shipped it in Twitter and it got viral.
0:13:01 It was like the first time that we
0:13:04 lived something that I had read about in terms of,
0:13:06 this is what PMF looks like.
0:13:09 That’s the first time I was like, oh, Jesus Christ.
0:13:11 OK, this is how it looks.
0:13:16 OK, I see you go to sleep and I can feel the heartbeat.
0:13:17 And then you sleep three hours.
0:13:21 You wake up because you know that there’s stuff broken.
0:13:23 The email starts to get flooded.
0:13:24 Twitter starts to go in there.
0:13:28 Suddenly, literally, every day was like crazier than the one
0:13:29 before.
0:13:31 I was like, oh, my God, 1,000 people.
0:13:32 Oh, my God, 10,000 people.
0:13:34 And there’s like, oh, my God, like,
0:13:36 Football Club Barcelona, like number one soccer club just
0:13:37 used us.
0:13:37 What?
0:13:40 And why isn’t like, how many followers they have?
0:13:42 Oh, my God, 100 plus million followers on Instagram.
0:13:46 OK, I feel like it was actually hard in the sense
0:13:52 of as a founder, you’re like, I’ve put multi amounts of years
0:13:54 into like many things.
0:13:56 And then the thing that we literally are like,
0:13:59 it’s not important gives you all the success.
0:14:01 So it’s a moment of reflection.
0:14:07 You’re like, sometimes like the world throws truths at you.
0:14:11 But those years of work were not all for nothing.
0:14:14 And I don’t think like the years that we’ve been working on
0:14:14 was like a waste.
0:14:15 No, it actually is.
0:14:18 Oh, that’s how you learn on the technical level, how it works.
0:14:20 I mean, because it was so much failure,
0:14:23 we learn about, OK, how do you communicate with your co-founder?
0:14:26 I think that’s something important to note about those times
0:14:30 is that we were very, very aware that this was a trend
0:14:35 and that this was not the end product that we were building.
0:14:38 It was like almost like a marketing engine
0:14:41 that we were using to get better branding and to get known.
0:14:43 They were finding us because of one reason,
0:14:45 but they were staying because of another one,
0:14:47 which was like this other product that we were working on.
0:14:51 Even when we knew that, I think that the core learning
0:14:56 that we got from this experience is that the AI field
0:14:59 changes constantly, like every month or every two months.
0:15:03 There are new breakthroughs, new techniques, new ways
0:15:04 of doing things.
0:15:07 And the tool that we were building,
0:15:09 it was like already starting to get too complex
0:15:12 because we were trying to put everything in a single tool.
0:15:14 And I think that what we learned with the experience
0:15:18 of the spiral virality is that there’s a lot of value
0:15:23 on simplifying super niche and simple use cases.
0:15:29 And that was the case again when LCMs were released.
0:15:32 We saw this technology, and at that moment,
0:15:35 we used all the experience that we got from the first virality
0:15:36 to engine the second one.
0:15:39 And the second one, we knew that it was not a trend.
0:15:41 It was something extremely valuable.
0:15:44 We were like finally being able to get that interaction
0:15:46 that we were looking for for almost years, right?
0:15:49 Like we can generate images in real time
0:15:51 and have full control of the colors, the composition,
0:15:52 the shapes, everything.
0:15:55 That was almost like a dream come true.
0:15:58 Victor and Diego have now hit virality several times over,
0:16:00 but can you engineer that momentum?
0:16:04 In some cases, it’s all about having a single critical feature
0:16:05 not offered elsewhere.
0:16:07 Mohamed from IDU Grimm.
0:16:12 So basically, it was the version 0.1, as we called it.
0:16:17 And this is back then in September of 2023.
0:16:19 And it was a model that was working.
0:16:26 It wasn’t perfect, but we felt like it’s already good enough
0:16:27 to give it to users.
0:16:30 And it was the first model that could put legible text
0:16:31 into images.
0:16:35 So it kind of went viral because of the unique capability
0:16:40 of the model, somehow the ability to put text into images
0:16:42 felt needed.
0:16:46 But in other cases, it’s about cleverly enabling the masses,
0:16:49 or in this case, the memesters, by drastically reducing
0:16:51 the barrier to participate.
0:16:54 Here’s hung with fake old story.
0:16:55 It went pretty viral, right?
0:16:58 What was that like experiencing to put a product
0:17:01 in the hands of so many users and also see that kind
0:17:02 of spread on its own?
0:17:06 Yeah, it was– we didn’t anticipate that for sure.
0:17:09 In the very beginning, we were thinking most targeting content
0:17:10 creators.
0:17:14 But somehow the meme makers and memesters stayed catch up on it.
0:17:16 And that’s how it got pretty viral.
0:17:19 And also, thanks to some of the templates,
0:17:22 we spent so much time discussing why this is the case.
0:17:25 There was this template, the Joker Lil Yachty coming
0:17:26 on the stage.
0:17:28 And there’s a Joker character that
0:17:29 we placed down on the video.
0:17:33 And we’ve seen millions of different characters
0:17:35 just remixing the same moment.
0:17:38 And we realized that the main reason was used to use.
0:17:42 It’s so easy to– basically, you can upload one image.
0:17:44 And then one click, choose that template.
0:17:47 And then in just a matter of seconds,
0:17:51 you’ll have yourself basically in that same moment.
0:17:54 Maybe one other aspect of the virality is, as you said,
0:17:56 the meme makers got ahold of it.
0:17:59 There’s this kind of fun, maybe even silly, aspect to it.
0:18:00 How have you thought about that?
0:18:03 Well, I think that speaks to the entertainment value.
0:18:07 And for anything to have real entertaining value,
0:18:07 it has to work.
0:18:09 It has to work well.
0:18:12 And that actually requires a lot of rigorous in the research
0:18:13 side.
0:18:16 So we are pretty serious about being silly.
0:18:20 And it takes quite a bit of rigorous research to do that.
0:18:22 And the second thing is, you have
0:18:25 to have a tool that provides precise control.
0:18:28 And then because people are getting what they want,
0:18:31 they can have all kind of a variety of fun with it.
0:18:33 You’ve mentioned characters and templates a few times.
0:18:35 What are some of your favorite examples
0:18:37 of those generated on the platform?
0:18:40 One is the Joker/Kamiyanthus state template.
0:18:42 That one is basically the moment we realize,
0:18:45 well, actually, people want to remix.
0:18:48 And there is this virality and memes aspect of it.
0:18:53 And the second one is, there has been one Rakuten advertising
0:18:53 song.
0:18:55 And people are dancing with this.
0:18:57 And we’re also seeing millions of people remixing
0:18:59 that same template.
0:19:02 And this is interesting for us, because you make us realize
0:19:06 that as long as there is this fun elements to it,
0:19:08 people actually don’t mind this content
0:19:10 having a little bit of a brand message.
0:19:11 When you think about applications,
0:19:14 and I know it’s early days, but have there
0:19:16 been any that have surprised you about the ways
0:19:19 that Vigil has been applied every time a founder creates
0:19:20 a product?
0:19:22 They have applications that they envision.
0:19:24 And then the best products are often
0:19:27 people are using them in alternate ways that surprise them.
0:19:30 Yeah, that was exactly the case for us.
0:19:32 In the very beginning, we were mainly thinking
0:19:34 of movie makers, game makers, using–
0:19:37 this might be a quick animation,
0:19:38 pre-visualization tool for them.
0:19:40 It’s actually pretty useful for that.
0:19:43 And we’ve also seen the early users adopting to that.
0:19:46 But then we never anticipated the meme search.
0:19:50 So since that, we’ll be also providing those templates.
0:19:53 So we’ve been keeping track of the latest trendy dance moves,
0:19:55 sports events, et cetera.
0:19:59 And we’ve also seen content creators hopping on to this.
0:20:01 They are actually reaching out to us,
0:20:04 say, can you feature our dance, our song, on your platform?
0:20:07 And then can we collaborate on promoting
0:20:10 some of those that’s been really interesting?
0:20:13 We ask Connor the same question around what he’s learning
0:20:16 by seeing how the masses are using UTO.
0:20:19 The model we originally launched was a model which
0:20:20 generated 32-second clips.
0:20:22 And so to make it kind of a full track,
0:20:24 you would extend that in various directions.
0:20:27 You would add an intro, add maybe a chorus and an outro
0:20:28 and stuff.
0:20:29 And you would build a song like this.
0:20:31 And you would start with these chunks.
0:20:33 And I suppose we’ve actually come to realize quite quickly
0:20:36 that people’s experience with music when they ask for, say,
0:20:38 a song is actually a lot more focused on that.
0:20:41 So they kind of want a song that begins at the beginning,
0:20:42 maybe ends at the end.
0:20:44 It’s maybe– it doesn’t have to be long.
0:20:45 It could be like a short two-minute clip.
0:20:47 But it has a verse and a chorus and a verse.
0:20:48 And there’s a structure to it.
0:20:52 And so I suppose we actually underestimated just how important
0:20:52 that was.
0:20:54 And so that’s something we were making steps to where
0:20:55 it’s rectifying recently.
0:20:59 11 Labs was also no stranger to the surprising and inspiring
0:21:00 user behavior.
0:21:03 Yeah, I think one of the most surprising ones
0:21:05 was people who had lost their voices
0:21:09 and then had used 11 Labs to bring their voices back to life
0:21:11 and then do the thing they loved doing.
0:21:13 So we had Lori Cohen, who was a lawyer.
0:21:15 She lost her voice one morning.
0:21:18 And a friend of hers helped her replicate her voice
0:21:19 with 11 Labs.
0:21:21 And then she was back in the courtroom delivering arguments.
0:21:24 And that, to me, is just such an incredible moment
0:21:26 because you don’t expect that.
0:21:28 And I think our idea was like, hey,
0:21:29 we’re going to give ideas of voice
0:21:31 with our product and our tools.
0:21:33 But this gave someone their own voice back.
0:21:35 And I think that was such an amazing thing to see.
0:21:38 And we saw that again with a climate activist, Bill
0:21:41 Wheel, who was delivering his award speech.
0:21:43 He suffered from ALS, unfortunately,
0:21:45 but, again, was able to replicate his voice
0:21:46 and then deliver that award speech.
0:21:49 So I think those kinds of things are just like–
0:21:51 you’re like, wow, technology being used in a way
0:21:51 we didn’t see it.
0:21:54 And now we want to lean into that and, of course, help others.
0:21:54 Yeah.
0:21:56 Maybe in the opposite sense, have there
0:21:58 been any applications that you’ve actually
0:22:00 built or designed for where you’re like, everyone’s
0:22:02 going to use it for this, obviously.
0:22:04 Or that’s actually not been the case?
0:22:04 It’s interesting.
0:22:07 When we launched dubbing and automated dubbing,
0:22:08 we thought, yeah, this is it.
0:22:10 Like, everyone’s just going to use automated dubbing.
0:22:10 It’s going to be great.
0:22:13 And of course, with dubbing, one of the most important things
0:22:14 is accuracy, right?
0:22:16 And so automated dubbing, we realized,
0:22:19 people still want a lot of creative control on that.
0:22:22 And so we ended up having to build dubbing studio, which
0:22:26 allowed people to go really fine tune that dub and change
0:22:27 a lot of the content.
0:22:29 And then we also introduced 11 Studios,
0:22:31 which was basically creative teams
0:22:34 that help you dub your content with professionals.
0:22:35 We’re really good at that.
0:22:37 And so we realized that actually was
0:22:40 what people needed more of and not just automate everything
0:22:41 and all the things, right?
0:22:43 And then it actually picked up again.
0:22:45 And this is something even when I was working at Palantir,
0:22:47 you learn, which is like the temptation
0:22:49 to try to automate everything or to use intelligence
0:22:50 for everything.
0:22:52 But actually, there’s so much value
0:22:54 in having someone in the middle and still having
0:22:57 that human touch to take it to that final step with something
0:22:59 we learned with dubbing.
0:23:02 And as these companies get all this new data,
0:23:03 it’s not always easy to figure out
0:23:06 who they should be catering to.
0:23:09 So how do you think about what you build and for who?
0:23:11 Your TAM is everyone, in theory.
0:23:14 I think what we acknowledge is that we probably
0:23:17 have different types of users, like distinctly different types
0:23:19 of users, at the very top being, does someone
0:23:23 in a studio who’s making an album at the very top level?
0:23:26 And then at the end of the scale is maybe someone
0:23:28 on their phone who wants, in a minute,
0:23:31 they want just a funny song to send to their friend.
0:23:34 And those are two very different experiences
0:23:38 and somewhat similar to the kind of output
0:23:41 you can get from just an instrument in general.
0:23:42 Someone can have a guitar at home
0:23:44 that they play just to have fun from time to time.
0:23:46 It’s like a totally personal thing.
0:23:47 It’s not anything necessarily serious.
0:23:50 It’s just a way to express yourself with it musically.
0:23:52 And the same way someone can take that same guitar
0:23:55 and a professional can take it into a studio
0:23:57 and make it part of something fantastic,
0:24:00 we like the technology to basically enable
0:24:02 all ends of the oil parts of that spectrum.
0:24:05 Several are unsprisingly using their flywheel of new users
0:24:08 to inform their decisions.
0:24:11 Yeah, we kind of use our user base
0:24:14 and the prompts that they enter into the system
0:24:18 to decide how to evaluate the quality of the model
0:24:20 and what to prioritize.
0:24:23 What’s interesting is our users used ideogram
0:24:25 to tell us what they want.
0:24:27 So they were like, we want image upload,
0:24:30 we want comment, we want more servers.
0:24:34 So I guess the good news is we already have this flywheel
0:24:36 of users coming and using it.
0:24:38 Some are paid, some are free.
0:24:40 And that sets the vision for us.
0:24:43 Hung from Vigol has actually used these new learnings
0:24:46 to expand who they’re building for.
0:24:49 How are you thinking about who you now build for, right?
0:24:51 Are you pivoting or adjusting
0:24:53 to incorporate these new use cases?
0:24:57 So we are broaden our target audience in this sense.
0:25:01 So we are seeing this as eventually we’re going towards
0:25:04 the direction of a new type of AI power content platform.
0:25:06 And the content platform is really important
0:25:08 to have all these creators.
0:25:11 And those are still the content creators, the artists,
0:25:13 the movie makers, the game makers,
0:25:14 the demo game designers.
0:25:18 They are the sources for all those new ideas,
0:25:20 all those new templates.
0:25:23 And then we’re broaden this into content consumers.
0:25:27 Basically Vigol is a new way to consume content.
0:25:29 Before AI, it was mainly like,
0:25:33 if I like the moment I will share it, I will like it.
0:25:35 But there’s a deeper engagement
0:25:36 you can have with that moment.
0:25:39 I can basically, I love this moment so much
0:25:42 that I want to put my own avatar in it.
0:25:43 It’s almost like in a parallel universe,
0:25:47 I want to see how this looks really if that moment myself.
0:25:49 So this is a new kind of content consumption.
0:25:53 And that’s actually one of the most important aspect.
0:25:58 The variety actually comes from all those creative ideas.
0:26:00 So for us, it’s all about empowering
0:26:02 those creative community first,
0:26:03 making sure they have what they want.
0:26:04 They have the best tool.
0:26:07 They have early access to new features.
0:26:09 They have almost private channels.
0:26:12 They have almost unlimited access.
0:26:13 The team at Krea, on the other hand,
0:26:16 is more focused than ever on experimentation
0:26:18 and their signal for success.
0:26:22 When your users are better at using your tool than yourself.
0:26:25 How I think about it is that every tool that we launch
0:26:27 follows a similar process.
0:26:30 And I think that it all starts with a hypothesis.
0:26:32 And I think that this initial hypothesis
0:26:33 needs to come from the founder
0:26:35 and needs to come from your own intuition.
0:26:38 But we are wrong a lot of times
0:26:40 in the way how we validate these ideas.
0:26:43 And when we are wrong, it’s through listening to the community,
0:26:45 seeing what they do with the tools.
0:26:47 And I think that a good rule of thumb
0:26:50 or something that I found that is a good north star
0:26:52 to realize when something is good or not,
0:26:56 is when your users are better at using your tool than yourself.
0:26:59 And that has been key to me, because with the real time,
0:27:01 I was seeing things that I was like, how the fuck?
0:27:02 Did they create that?
0:27:04 And same thing with the video tool.
0:27:06 Like with the video tool, I was trying to do a demo,
0:27:08 like trying to showcase cool stuff.
0:27:10 And I was trying things, and I was not getting there.
0:27:12 And I was looking at Twitter at all the things
0:27:14 that our users were creating with our product.
0:27:16 And I couldn’t get to that quality.
0:27:18 I couldn’t get to those results.
0:27:20 So I think that every time that your users are using
0:27:23 your product better than what you are, that’s a good sign.
0:27:26 – Meanwhile, Justin and Maxis of it are charting new ground,
0:27:29 but also figuring out new limits.
0:27:31 – Stable diffusion allowed you to make anything.
0:27:34 And so when we launched, I wanted to make sure
0:27:37 that we could continue to support that community.
0:27:38 But it was so diverse.
0:27:42 And there’s running meme of things you can make
0:27:43 with stable diffusion.
0:27:45 And in the front is like somebody making funny memes.
0:27:48 And then there’s a train coming that’s porn, right?
0:27:50 Sure, people know that you can make all of this stuff.
0:27:53 I mean, that’s the point of this tech, make anything, right?
0:27:55 And so it was important for us to say,
0:27:58 hey, we want to be able to support this tech as it develops.
0:28:00 It means that we need to embrace all of it.
0:28:01 And that’s not easy.
0:28:06 It’s been incredibly difficult to set up policies
0:28:08 that allow the creation of all things
0:28:13 in a way that’s not going to hurt people
0:28:16 and to also do it in a way that makes it so that people
0:28:18 still have the level of control that they need
0:28:22 to prevent the creation of content that can’t be there.
0:28:23 – In the beginning, our policies were very straightforward.
0:28:26 They were kind of just like, look, as long as it’s not illegal
0:28:29 and as long as it’s not just ethically, completely debased,
0:28:31 then we’ll let it on the platform.
0:28:35 And we were okay when we had the small enough user group
0:28:37 with kind of leaving it even like that vague.
0:28:40 We found over time that we’ve had to really kind of specify
0:28:42 ’cause it turns out that there are just like subsections
0:28:44 of the internet that are into just the absolute strangest
0:28:46 things you’ve never heard of at all,
0:28:48 which can be really funny, which can be really cool.
0:28:49 And some of that is really interesting.
0:28:51 And some of it is just, oh my gosh.
0:28:54 And it’s like a balancing act of figuring out, okay,
0:28:56 what are, you almost have to grow as a person.
0:28:59 And we created like a council of moderators around here
0:29:01 to on our platform to really kind of like get together
0:29:04 and look at when these new things pop up and be like,
0:29:05 how do we feel about this?
0:29:06 One of the things that really blew my mind
0:29:08 when we were getting into the whole moderation aspect was,
0:29:10 like, oh, we’ll just do what other platforms do.
0:29:11 We’ll do what Imager does, we’ll do what Reddit does.
0:29:13 We’ll just copy kind of like what they’re doing.
0:29:16 And as we kind of dug into what they do,
0:29:19 is they don’t define any of this.
0:29:20 None of this is defined.
0:29:22 We had to come up with terms of how do you define
0:29:23 what a child is?
0:29:25 How do you define what is photorealistic?
0:29:27 How do you define what is and isn’t all of these terms?
0:29:31 But before, really didn’t have any really set definition.
0:29:32 – Perhaps it shouldn’t be surprising
0:29:34 that there are new moderation challenges
0:29:36 since this industry is so fresh
0:29:39 with new ideas coming from a new breed of creatives.
0:29:41 In fact, we heard about this range
0:29:43 in both prosumers and professionals
0:29:45 from most of the founders we spoke with.
0:29:46 Here’s Victor from Korea.
0:29:50 – The range of creatives is quite wide.
0:29:52 Like the kind of people that use Korea
0:29:56 can come from having 20 years of working
0:29:58 in the creative industry and being like,
0:30:00 I don’t know, three the artist
0:30:02 or people doing graphic design
0:30:05 or even photographers or these kind of people.
0:30:07 But we also find a lot of folks
0:30:10 who don’t have a professional creative background.
0:30:11 For the professional ones,
0:30:14 you can find them doing a lot of prototyping.
0:30:17 Like for example, when they start working on a new project,
0:30:21 they may go to Korea to really quickly brainstorm
0:30:22 some ideas that they have
0:30:24 and they would use the real time tool
0:30:26 that we have for that.
0:30:29 And they can do like a very simple sketch
0:30:30 add a text from
0:30:32 and have something that looks super realistic
0:30:34 and that can either give them ideas
0:30:37 and maybe even serve as a final deliverable
0:30:38 depending on what they’re doing.
0:30:41 And when we’re talking about like a less professional creative,
0:30:43 it’s honestly more about having fun.
0:30:46 And they are using Korea for everything that you can imagine
0:30:51 from imagining new walls to creating paintings
0:30:54 to creating like characters or all sorts of things.
0:30:55 – And as more participate,
0:30:57 these new platforms generate new talent
0:31:02 but also new expectations like expectations in speed.
0:31:06 – And on the meantime, what we’re doing is building community
0:31:09 and bringing to the community what they want now,
0:31:11 just focusing on what can we do now
0:31:13 with the technology that is out there.
0:31:17 We are very deep into AI communities
0:31:20 and every time that there’s something
0:31:22 that we think that is valuable from a creative point of view,
0:31:24 we go ahead and we execute it very, very fast.
0:31:27 So the way how we’re working is almost like a video game company
0:31:30 where instead of video games, we are building tools
0:31:33 and every six months or so, there’s a new tool
0:31:35 because the space just happened to evolve in a way
0:31:38 that every six months, there’s a new technology
0:31:40 that you can use in order to make a new tool.
0:31:42 And that’s gonna keep being like that until we get to these
0:31:44 like real-time multimodal systems
0:31:47 that allow us to do something way more interesting.
0:31:51 – This new wave has also shifted people’s willingness to pay.
0:31:53 Back to Anish.
0:31:54 I think the willingness to pay
0:31:57 and the amount that consumers are willing to pay is really high.
0:31:59 And that’s really interesting because for so long,
0:32:02 we’ve had these sort of patronage models
0:32:03 for how to fund the arts.
0:32:05 And there’s been this belief
0:32:08 that there’s a sort of decreasing interest in paying for art.
0:32:09 And instead, we’re seeing the exact opposite.
0:32:12 People want to pay for art and pay for tools to make art
0:32:13 and pay a lot.
0:32:17 So that’s a really, really exciting development to me.
0:32:18 – And this willingness to pay
0:32:21 is also unlocking new business models.
0:32:22 – People make so many things
0:32:24 because it’s a tool for creating anything.
0:32:27 And to see the things that people can create,
0:32:29 whether that’s assets for a game
0:32:31 or videos of flowers that are dancing,
0:32:32 it’s just endless.
0:32:33 The possibilities are endless.
0:32:35 And it’s inspiring to see how people
0:32:36 are kind of pulling it to do new things.
0:32:38 – That was Justin from Civet,
0:32:40 which is also working on a new way
0:32:43 to reward AI artists for their contributions.
0:32:45 – When we were getting this going
0:32:47 and we were really like realizing this could be a business,
0:32:49 was we interacted with a lot of the people
0:32:49 who are doing this creation.
0:32:51 And it’s a lot of time and it’s a lot of money.
0:32:52 That’s a lot of like technical skill
0:32:54 that goes into making these things well.
0:32:55 And people were doing it,
0:32:56 thousands of people were doing it,
0:32:57 just for the love of the game.
0:32:59 Like they just really enjoyed the clout
0:33:01 and the entertainment factor.
0:33:02 They got their position on the leaderboard.
0:33:05 – The leaderboard, oh my God, the leaderboard.
0:33:07 And it became pretty clear that,
0:33:09 look, this is almost like a whole new creator economy
0:33:09 that can come out of this
0:33:11 because it’s a group of people who are putting effort
0:33:14 and love into something that could very easily
0:33:15 become livelihoods for them
0:33:17 if they had even the smallest way to monetize it
0:33:18 based on the number of eyes they’re getting
0:33:19 and uses they’re getting.
0:33:21 So yeah, a very clear goal
0:33:22 from the very beginning was like,
0:33:24 let’s figure out how we can keep the creators monetized
0:33:26 while maintaining the open source ethos.
0:33:28 – We actually just announced something
0:33:30 that we’re hoping to roll out over the next six weeks.
0:33:31 Let me give you a little bit of history.
0:33:33 So we launched a creators program four months ago
0:33:35 and we opened it to a small cohort
0:33:37 of essentially 50 creators.
0:33:39 We opened applications and took essentially people
0:33:41 that met certain criteria
0:33:43 and have been experimenting with ways
0:33:45 that we can help them monetize their work.
0:33:48 What we’ve landed on for this next generation
0:33:50 that we’re hoping to open up in these next six weeks
0:33:53 is making it so that people can earn for the generation
0:33:55 that people are doing on our site.
0:33:56 So if they make a resource
0:33:59 that’s intended to produce a new character,
0:34:01 like a consistent character that they’ve made
0:34:03 and somebody chooses to use that in the generator,
0:34:05 then they’re gonna get their share of 25%
0:34:07 of what we charge for that generation.
0:34:10 So the aim is to make it so that these people have a way
0:34:14 to get essentially paid for allowing the convenience
0:34:15 of using their resource on our site.
0:34:17 – One of the main things that we saw right away
0:34:19 before we had the time to be able to implement
0:34:21 any real monetization stuff for creators was,
0:34:23 we put in a DM system simply because we knew
0:34:25 that there’s a lot of people who are contacting creators
0:34:26 for work outside of the platform.
0:34:28 And because of that, I mean,
0:34:30 we just get untold number of people contacting us
0:34:31 being like, “Thank you so much as platform.”
0:34:32 Because of that, I was able to get hooked up
0:34:34 with Hugo Boss or Hyundai
0:34:35 or some of these other people
0:34:36 who are suddenly using this technology.
0:34:38 And it’s completely changed my life
0:34:40 before I was making $30,000 a year as a waiter or whatever.
0:34:44 And now I’m making six figures doing this whole new thing
0:34:45 that’s a passion for me.
0:34:46 And I have lost count
0:34:48 on the number of people who’ve contacted me about that.
0:34:49 So it’s really cool to see.
0:34:51 So from a services side, we want to kind of enable that
0:34:53 and make it even easier for people to be able to sell
0:34:55 their services, their expertise,
0:34:57 directly to businesses from the system.
0:34:58 – It’s not alone here.
0:35:02 11Labs is also building a marketplace for voices.
0:35:06 – I know you guys are building kind of a marketplace of sorts.
0:35:08 So people can upload voices
0:35:10 or they can use voices that others have uploaded.
0:35:12 – Yeah, I think it’s a really exciting way
0:35:15 to give folks a way to earn passive income as well.
0:35:16 Maybe you were a voice actor
0:35:18 and you weren’t getting the gigs you wanted,
0:35:20 but now you can put your voice out there
0:35:21 and you might become extremely popular.
0:35:25 And we’ve seen people earn quite well on our platform.
0:35:28 And so the library is just a great way
0:35:30 to one, put your content out there.
0:35:32 And we want to partner with more voice actors, honestly,
0:35:34 to have more expressive voices
0:35:37 and then give people great voices to create content with.
0:35:39 So two-way street.
0:35:41 – But it’s not just the marketplace.
0:35:43 It’s also the interface.
0:35:46 – I think we’ve always had the dream of voice interactions
0:35:47 with all our products.
0:35:49 If you think about Star Trek
0:35:51 and Knight Rider talking to his car kit,
0:35:52 it’s something that’s been a part
0:35:53 of pop culture history forever,
0:35:57 but I don’t think we’ve had the quality and the sound
0:36:00 and for it to feel as natural as it should have been.
0:36:02 And so I think we’re getting to that point
0:36:04 where the interactions between large language models
0:36:08 using voice interfaces is becoming incredibly natural
0:36:10 and feels like talking to a person.
0:36:11 And so I do totally see a future
0:36:14 where a lot of this physical interface
0:36:16 that you’re tapping around with is going to just fade away
0:36:18 and you’re going to be able to ask the questions
0:36:21 you want to ask and have the conversations you want to have.
0:36:24 I know her is the hot topic movie of the AI space,
0:36:25 but I think there was one thing in that movie
0:36:28 that stuck with me more than just the interactions
0:36:29 he was having with her,
0:36:31 which was there was this scene in the movie
0:36:34 where everyone was down looking at their phones
0:36:35 and kind of scrolling.
0:36:37 And there’s this inflection point somewhere later
0:36:39 in the movie where actually everyone’s kind of talking
0:36:41 to something in their ear.
0:36:44 And I think that is a very precious take that they had.
0:36:46 And I think we’re going to see more of that.
0:36:48 It’s just going to be natural conversations
0:36:51 we’ll be having with this AI or any interface.
0:36:53 – Yeah, it reminds me of my husband’s grandmother,
0:36:55 says that the first time she ever heard someone talking
0:36:58 on a phone in the grocery store,
0:37:00 she thought they were talking to themselves.
0:37:02 Because all of these new interactions, right?
0:37:03 You’re just not used to,
0:37:05 or the people who go to prison and come out
0:37:07 and 10 years later, they’re like,
0:37:09 why is everyone looking down?
0:37:11 And then I realize that we have these crazy computers
0:37:11 in our pockets.
0:37:12 – Totally.
0:37:14 The thing that I love about AI in particular
0:37:16 and all these AI creative tools is
0:37:19 the magic is you had an idea
0:37:20 and now you can imagine it, right?
0:37:22 You can imagine the image that you wanted
0:37:24 and that was in your head and the dream you had.
0:37:26 And now we’re saying you can imagine the sound
0:37:27 that you’re probably hearing in your head
0:37:29 that no one else can hear yet.
0:37:31 – But it’s not just a new UI.
0:37:34 Perhaps it’s a new approach to modeling the world itself.
0:37:36 Hang from Viggo.
0:37:38 – One thing I really look forward to is,
0:37:40 like I said, the next generation of the model.
0:37:44 So we are really hoping to extend this character model
0:37:48 to more the rest of the world, like objects and the scenes.
0:37:52 And so I think those are two general passes
0:37:54 towards modeling the real world.
0:37:58 One is more on, we’ve seen this pixel level approach.
0:38:00 So diffusion models are really good at that.
0:38:02 But it has this drawback of,
0:38:04 it’s really hard to manipulate pixels.
0:38:08 And the real world is essentially, is really, is physical.
0:38:11 So pixel is not really a efficient representation for it.
0:38:14 But it has the advantage of you can train with any video
0:38:16 and it generates anything.
0:38:20 And the hope there is, if we scale it up to a certain extent,
0:38:22 a controllability will kind of emerge.
0:38:25 But we’re taking another kind of different path
0:38:28 in that we want to nail down better the first,
0:38:32 making sure it’s just as precise, as controllable,
0:38:35 as a graphics engine, and then we scale up from there.
0:38:39 So I think this, how those two passes evolve
0:38:44 and how actually they can be combined into one immersive experience.
0:38:46 As we close out this episode, it’s hard to understate
0:38:50 just how much these tools are shifting, what it means to be creative.
0:38:54 To both existing artists and to those who never would have called themselves artists before.
0:38:56 Connor from UDO.
0:39:00 The threshold for someone going into a studio and recording something like that
0:39:01 was way too high.
0:39:04 Whereas now, the promise of the technology is that
0:39:06 it brings an order of magnitude or two orders of magnitude,
0:39:09 more people into the creative kind of experience, right?
0:39:13 Like, people can express themselves in this way,
0:39:16 but kind of even more concretely, as moments happen in the world,
0:39:19 different cultural moments, you can attach music to them now.
0:39:23 Because it can be dynamically attached to these things in interesting ways.
0:39:24 And this is super compelling.
0:39:27 This is a kind of a market that didn’t really exist before
0:39:30 just because it wasn’t actually possible to explore this way.
0:39:33 I think as well as that, we’ve been fascinated with how
0:39:37 at the top level, say with the existing artists or existing producers,
0:39:40 how this can basically work as an ideation machine,
0:39:45 like a kind of well of infinite creativity that you can just pull from for ideas.
0:39:47 Maybe you have the beginning of a track, you have a riff, you have a beat.
0:39:50 You want to see where could this go from here?
0:39:52 If I remix this a bit, what are variations on this?
0:39:54 And that’s a super compelling thing to do as well.
0:39:58 Again, because it’s something that before took a lot of time.
0:40:02 And so it just accelerates the creative experience for professionals like that as well.
0:40:05 I have yet to meet an artist who’s actually used the products
0:40:07 that is worried about the products competing with them.
0:40:11 The biggest worry that I hear over and over is that somebody is going to take them away.
0:40:16 Diego from Korea with a great reminder of just how monumental this shift is.
0:40:21 I was a creative myself doing graphic design, photography.
0:40:25 I even tried to make video games in flash, motion graphics in After Effects,
0:40:30 digital sculpture in Zbrush, 3D modeling for architecture visualization.
0:40:35 And I was like, it’s almost like I felt the fear of,
0:40:38 “Hey, what’s the point if this thing can’t do everything,” right?
0:40:40 But I don’t think that’s the case.
0:40:49 What I think is happening is that we’re just giving so much power to creatives
0:40:56 that things that were like a job in a way like now you don’t even think about them.
0:40:58 That’s what technology does, right?
0:41:03 Like one day it is a lifetime work to move from the east coast of the US
0:41:06 to the west coast and people die on the process.
0:41:11 So now you’re like, “Oh, it took me like 20 minutes at the line to get to the airport thing.”
0:41:15 And you don’t even think about the fact that you flew like a great god through the planes.
0:41:17 Instead, you’re just thinking at a higher level.
0:41:23 You’re just, I don’t know, flying between coasts to make like bigger things, right?
0:41:26 So I feel like the same is going to happen.
0:41:31 Suddenly like coloring 3D models through texture things and all these repetitive things
0:41:37 like sketching and whatever like you will save so much time like of your life.
0:41:42 Because of not having to do that, you can focus on having even better and crazier ideas.
0:41:49 So I’m really, really, really excited to see what the creatives are going to be able to do.
0:41:51 All right, that’s all for now.
0:41:55 The demos shared during the day were followed by a gallery party at night
0:41:59 showcasing many of the artist’s work of the broader New York City creative community.
0:42:02 So if you want to get up close and personal with these tools,
0:42:07 head on over to a16z.com/aiart to check out their demos and more.
0:42:10 We’ll leave you with a little sneak peek.
0:42:17 Ladies and gentlemen, I am thrilled to be here at the a16z artist retreat.
0:42:19 Yeah, it gets you pumped.
0:42:20 So pumped.
0:42:21 So pumped.
0:42:24 You can generate whatever you want.
0:42:25 That was amazing.
0:42:28 Yes, so it’s a whole body swap.
0:42:29 Wow, this is so good.
0:42:33 We’re also working on a new type of memes.
0:42:36 Actually, I think that is better if we see it in slow motion.
0:42:37 All right, you should come see this.
0:42:40 This changes from being deterministic to being totally random.
0:42:42 Remix.
0:42:49 If you liked this episode, if you made it this far, help us grow the show.
0:42:52 Share with a friend or if you’re feeling really ambitious,
0:42:58 you can leave us a review at ratethespodcast.com/a16z.
0:43:03 You know, candidly producing a podcast can sometimes feel like you’re just talking into a void.
0:43:07 And so if you did like this episode, if you like any of our episodes,
0:43:08 please let us know.
0:43:18 We’ll see you next time.
0:43:27 [BLANK_AUDIO]
0:00:06 On June 27th, our team headed to New York City.
0:00:08 We are at the A16Z office
0:00:11 for the first ever AI artist retreat.
0:00:15 That was A16Z consumer partner, Justine Moore.
0:00:16 Justine was one of many partners
0:00:18 who attended this retreat,
0:00:19 which brought together the builders
0:00:22 behind some of the most popular AI creative tools
0:00:23 in existence.
0:00:28 That is, 11 labs, Korea, FIGL, Udio, ideogram, and civet.
0:00:31 All together with 15 top artists.
0:00:34 These are the folks who are often
0:00:36 doing the coolest things with these sorts of tools.
0:00:38 They’re kind of pushing the boundaries
0:00:40 of what the tools can create.
0:00:43 Today, you’ll get to hear from many of these AI founders,
0:00:45 who together with these artists
0:00:48 are advancing what it means to be creative.
0:00:50 Art is going to get better than ever.
0:00:53 The average art output is going to improve,
0:00:55 but so is the ceiling.
0:00:57 It also is a higher participation rate.
0:00:59 Everyone who’s interested in creativity
0:01:01 can be creative and express themselves,
0:01:02 which is just so cool.
0:01:04 That was Anish Acharya,
0:01:06 general partner on the consumer team,
0:01:08 but that’s not all.
0:01:09 I’ve been a founder twice.
0:01:12 I’ve been spinning records as a DJ for 25 years,
0:01:14 and I’m all about AI and art.
0:01:17 So what happens when you put all these investors,
0:01:20 leading artists and creative tool founders,
0:01:21 all into the same room?
0:01:24 I mean, the vibes have been immaculate.
0:01:25 And I think that the thing that’s the most surprising
0:01:27 is how much everyone has in common.
0:01:29 Like the founders are more creative,
0:01:33 and the creatives and artists are more technical.
0:01:33 I think the other thing
0:01:36 has just been how interdisciplinary it all is.
0:01:37 People making video,
0:01:39 want to play with generative audio,
0:01:40 people making music,
0:01:42 want to play with sound effects.
0:01:44 It’s just incredible to see.
0:01:45 One of the coolest things was
0:01:48 a lot of the founders had recognized people
0:01:50 by their online screen names or knew,
0:01:52 “Oh my gosh, you used my tool
0:01:55 to create this incredible song that went super viral.”
0:01:57 Or, “You used my product to make this
0:01:59 kind of amazing video animation
0:02:01 that our whole team was talking about for a week.”
0:02:03 These are people who have been interacting with each other
0:02:08 often daily online for the past six, 12, 18 months,
0:02:09 sometimes even two years,
0:02:13 but didn’t even know what each other looked like in person.
0:02:16 Now today, you get a behind the scenes look into this event,
0:02:19 including the origin stories behind many of these tools,
0:02:22 which by the way, some have never been shared publicly,
0:02:24 and have these tools,
0:02:26 which have all gone through their own viral moments,
0:02:28 are navigating this AI wave,
0:02:30 and what they see on the horizon.
0:02:31 Let’s get started.
0:02:35 As a reminder,
0:02:38 the content here is for informational purposes only,
0:02:40 should not be taken as legal, business, tax,
0:02:41 or investment advice,
0:02:43 or be used to evaluate any investment or security,
0:02:46 and is not directed at any investors or potential investors
0:02:48 in any A16Z fund.
0:02:50 Please note that A16Z and its affiliates
0:02:51 may also maintain investments
0:02:54 in the companies discussed in this podcast.
0:02:56 For more details, including a link to our investments,
0:02:59 please see a16z.com/disclosures.
0:03:06 Here we are in 2024.
0:03:08 We’re at an exciting inflection
0:03:10 where your creativity is being unbounded
0:03:12 by the tools available.
0:03:15 – I mean, we’re early, but there’s more people
0:03:17 making more art and more people making more tools
0:03:20 to make art than ever before.
0:03:23 And if you kinda look at the history of technology and art,
0:03:25 every single time there’s been a new technology,
0:03:28 the amount of art has dramatically increased.
0:03:31 People worry that drum machines would compete with drummers,
0:03:33 and instead there’s more people making more music
0:03:35 with both drummers and drum machines than ever before,
0:03:37 so I think there’s a sort of equivalent moment
0:03:39 here in technology and art,
0:03:42 where we’re at the beginning of everybody
0:03:45 who has taste and interest in art being able to make it.
0:03:48 – Many have drawn parallels to prior computing waves,
0:03:50 but is this any different?
0:03:51 – Well, what’s different is for the first time
0:03:53 we’re creating these sort of left brain things.
0:03:54 You know what I mean?
0:03:56 Computers and computing platforms
0:03:58 have really been in the business of precision,
0:04:00 and now we’re creating products
0:04:03 that are intentionally imprecise, beautifully imprecise,
0:04:06 so it just feels like a whole different flavor
0:04:09 for products and product design than we’ve ever seen before.
0:04:11 – So let’s introduce you to some of the people
0:04:12 behind those products.
0:04:14 – We have companies here covering basically
0:04:18 every sort of creative modality, image, video, music,
0:04:20 3D speech, all those sorts of things.
0:04:21 – That includes–
0:04:23 – Connor, I’m a co-founder at UDO.
0:04:24 – And–
0:04:26 – Omar, I’m the head of design at Love & Labs.
0:04:28 – Both companies are focused on audio,
0:04:30 with UDO focused on music,
0:04:32 while Love & Labs is tackling everything
0:04:34 from voice to sound effects.
0:04:35 Meanwhile, founders like–
0:04:39 – Mohammed, I’m the co-founder CEO at Ideogram.
0:04:40 – Victor and–
0:04:41 – Diego.
0:04:42 – Who are the co-founders of–
0:04:42 – Greya.
0:04:43 – And–
0:04:45 – Hong, I’m working on VEGLE.
0:04:46 – These founders are building
0:04:48 at the increasingly sophisticated world
0:04:51 of 2D imagery and video, plus 3D.
0:04:54 Ideogram, for example, lets you generate AI imagery
0:04:56 with accurate text embedded,
0:04:59 a surprisingly difficult technical feat.
0:05:00 VEGLE, on the other hand, is building
0:05:03 at the intersection of video and 3D.
0:05:06 Meanwhile, Greya’s come up with a suite of AI tools,
0:05:08 like upscalers and real-time generation.
0:05:12 Or, in the case of Civit, a new breed of marketplace.
0:05:14 – My name’s Maxwell Holker.
0:05:17 I am COO at Civiti and co-founder.
0:05:18 – Yes, and I’m Justin Mayer.
0:05:21 I’m the COO and co-founder of Civit as well.
0:05:22 – And CPO and CTO and–
0:05:24 – Lots of things, the joys of a startup.
0:05:26 We are a massive community of people
0:05:29 making tons and tons of AI creations,
0:05:33 using community-made models with community-made patches
0:05:35 to those models called LORAS.
0:05:37 We give people the ability to either train
0:05:40 on a few specific models, so a model focused on anime,
0:05:42 or a model focused on being semi-realistic.
0:05:44 Or they can select their own custom model
0:05:46 to train on top of.
0:05:48 – With AI moving so quickly, it’s clear
0:05:50 that we no longer live in a world
0:05:52 of just chat GBT and mid-journey.
0:05:55 Numerous companies have springboarded into the zeitgeist
0:05:57 and grown at unprecedented rates.
0:06:00 So we thought it was fitting to take a step back
0:06:02 and document this whirlwind of a journey.
0:06:04 While many of these founders have been quietly working
0:06:08 in research for years, their origin story often started
0:06:10 from scratching their own itch.
0:06:11 Muhammad from I8Gram.
0:06:15 – I guess part of it is that there is this thesis
0:06:19 that everybody has an innate desire to create.
0:06:24 And as humans, we have this inner creative child.
0:06:28 The education system sometimes kills
0:06:30 this creative child, unfortunately.
0:06:34 And what’s finally possible with technology and AI
0:06:38 is to help people express themselves visually
0:06:39 and creatively.
0:06:40 So that’s the interesting part.
0:06:45 When you think of using image for communication,
0:06:48 then you can communicate much more effectively
0:06:51 if you have image and text together.
0:06:54 – For Muhammad, it really was this unique combination
0:06:56 of text and imagery.
0:06:59 – For me, image and video is dear to my heart
0:07:01 and very personal.
0:07:03 – But for Connor, it was his connection to music.
0:07:06 – Music for me, I think, is very special medium.
0:07:07 It’s everywhere at all times.
0:07:09 Like it’s in the background when you’re at a restaurant
0:07:11 or a cafe, you’re listening to your headphones
0:07:12 and you’re going to work in the morning.
0:07:15 It really has an emotional resonance with people.
0:07:17 And for me, making that abundance,
0:07:19 like the kind of promise of generative modeling
0:07:21 is that a lot of this can be far more abundant
0:07:23 than it ever was before.
0:07:24 – And for Victor, it was his discovery
0:07:27 that programming itself was the creative gateway.
0:07:29 – When I discovered about programming,
0:07:32 that was great to me because I realized that through coding,
0:07:33 you can also be super creative.
0:07:37 But the moment where I discovered about early gen AI models
0:07:40 like DCGan and later on, StyleGan,
0:07:41 that’s when my mind was blown.
0:07:45 And when I realized about the creative potential
0:07:46 that this technology had,
0:07:48 and that’s when I fell into the rabbit hole.
0:07:49 And I feel like Korea, to me,
0:07:52 it’s been kind of the snowball that it has started
0:07:55 with me realizing that you can use artificial intelligence
0:07:57 in a creative way.
0:07:59 – But for Omar, it was building his own side projects
0:08:01 and a desire to share what he was learning
0:08:04 that actually propelled him into his role at 11 Labs.
0:08:07 – It’s really funny, actually.
0:08:10 I, over the last maybe couple of years,
0:08:13 started diving into AI tools when ChatGBT came out
0:08:15 and started making things on the side for fun.
0:08:17 One of those was a children’s book
0:08:19 that ended up accidentally going viral.
0:08:22 And that kind of was my journey into AI.
0:08:23 Through that and making that book,
0:08:25 I started exploring other AI tools.
0:08:29 And what I really enjoyed was sharing what I was doing
0:08:30 and how I made it.
0:08:33 So I discovered 11 Labs and made a podcast with 11 Labs
0:08:36 where I was talking to a fictional figure
0:08:38 and we were having a back and forth conversation
0:08:40 that also kind of did the numbers on Twitter.
0:08:42 And then I was like, I love using this tool.
0:08:45 I’m gonna make my own AI short movie.
0:08:46 Actually, Justine and I are friends.
0:08:48 And so I showed it to her and I was like,
0:08:51 I kind of need free credits because this movie
0:08:53 is using up all the credits on 11 Labs.
0:08:55 And she’s like, you should meet the founder, Maddie.
0:08:57 We met, we really hit it off.
0:08:59 And Maddie, in classic Maddie fashion,
0:09:02 was very direct and at the end of the call was like,
0:09:05 hey, we’re actually looking to hire someone to lead design.
0:09:06 Are you interested?
0:09:09 And then to work on a product that I’d used for over a year.
0:09:10 That experience also gave Amar
0:09:13 a taste of just how quickly this space moves
0:09:15 and also a hit of virality.
0:09:18 It happened because a friend of mine had their first kid
0:09:21 and I read her children’s book actually.
0:09:21 I was reading it and I was like,
0:09:23 this story kind of makes no sense.
0:09:25 So, so I went back home.
0:09:26 I’d been using the journey a lot,
0:09:28 chat to you two weeks old,
0:09:30 combine the two to create that book.
0:09:33 And then I was like, how do I get this published?
0:09:35 And Amazon has this amazing publishing service.
0:09:37 You can get a book out within 48 hours.
0:09:41 I had a paper back in my hand in 72 hours, so fast.
0:09:44 And it’s really interesting because writing a book
0:09:46 and publishing on Amazon is like,
0:09:48 it was almost like iterating on software.
0:09:49 If I discovered a type or whatever,
0:09:51 I just updated the PDF and the new book was out
0:09:53 and a new publishing line was out.
0:09:56 And so, yeah, I put it out there,
0:09:57 got a ton of virality from that.
0:10:01 And yeah, that was a really interesting experience.
0:10:04 – Pre-AI, we were in this era of consumer
0:10:07 where it was just really hard to get people’s attention,
0:10:09 really hard to get them to download a new app
0:10:10 or try a new tool.
0:10:13 You had to spend a lot of money on customer acquisition.
0:10:16 Now, just with the real excitement around AI,
0:10:18 if you make a cool product,
0:10:20 you can get it into the hands of people
0:10:22 and get them using and talking about it.
0:10:25 – This was the case for Victor and Diego at Cria,
0:10:28 who eventually met their own viral moment,
0:10:29 although it didn’t come easy.
0:10:32 – First of all, it was called Geniverse,
0:10:36 coming from Generative Universe, best name ever.
0:10:38 And essentially, it was like two things.
0:10:40 It was on the one side, an open source library
0:10:43 that it was kind of integrating all the cool stuff
0:10:45 that it was available at that moment.
0:10:47 And on the other side, it was a creative tool.
0:10:49 And the way how it looked, it was like super experimental.
0:10:52 Like, we didn’t really know how to do UI design
0:10:53 or any of that.
0:10:56 Like, the background really had stars and everything.
0:10:59 So it was like a galaxy, like the Generative Universe, right?
0:11:03 And then you could put text, you could put images,
0:11:05 and you had a few things that you could tweak
0:11:06 and you could generate images.
0:11:09 And you would see like the image evolving in real time.
0:11:12 And the images that you liked, you could keep them.
0:11:14 And they were added to this kind of universe.
0:11:17 And essentially, you ended up with a ton of images
0:11:18 in this interactive space.
0:11:22 So for us, it was always with the same idea in mind,
0:11:24 on the one side, controllability.
0:11:26 And on the other side, intuitiveness.
0:11:30 Like, how do we make tools that doesn’t look daunting?
0:11:32 Because AI, in the end, is like a new creative medium.
0:11:34 A lot of people are using it for the first time.
0:11:37 And we want them to have the inexperience
0:11:39 where the AI does what you expect to do.
0:11:42 And you don’t need to learn about crazy prom and generings
0:11:46 and all these tweaks up to get good results.
0:11:47 And on the other side is controllability
0:11:49 because we are dealing with creatives.
0:11:52 We are dealing with folks who are not just OK
0:11:54 with having a beautiful image.
0:11:55 They want that beautiful image.
0:11:59 So these are the two core principles that we had since then.
0:12:03 And we built kind of a Figma-ish interface for AI.
0:12:07 And we had every single utility that you had at that point
0:12:08 with the stable diffusion in there.
0:12:11 We had like thousands of AI models that you could use.
0:12:13 We had every single technique, like every control net.
0:12:15 Everything was in there.
0:12:16 But you know, it was not working.
0:12:18 It was like a learning curve that some people
0:12:20 were just not willing to take.
0:12:24 So then we have the first kind of virality moment
0:12:27 when we ship this thing that it was almost like an equivalent
0:12:28 to a meme generator.
0:12:31 I remember that we were seeing all of these images on Twitter
0:12:33 with the spirals, right?
0:12:35 So we were like, well, what’s going on with these spirals?
0:12:36 Like, we can do it.
0:12:38 This is like one day of work.
0:12:40 And I remember at that point, Diego was like,
0:12:41 we should do something with this.
0:12:43 We should do something with this.
0:12:44 It’s getting so viral.
0:12:45 And I was more like in the mood of, like,
0:12:48 we need to ship this whatever feature we were working on.
0:12:50 At that moment, until at one point, we were like,
0:12:51 OK, let’s fucking do it.
0:12:53 And we did it like in the sketchiest way possible,
0:12:55 like in one or two days.
0:12:57 And we shipped it in Twitter and it got viral.
0:13:01 It was like the first time that we
0:13:04 lived something that I had read about in terms of,
0:13:06 this is what PMF looks like.
0:13:09 That’s the first time I was like, oh, Jesus Christ.
0:13:11 OK, this is how it looks.
0:13:16 OK, I see you go to sleep and I can feel the heartbeat.
0:13:17 And then you sleep three hours.
0:13:21 You wake up because you know that there’s stuff broken.
0:13:23 The email starts to get flooded.
0:13:24 Twitter starts to go in there.
0:13:28 Suddenly, literally, every day was like crazier than the one
0:13:29 before.
0:13:31 I was like, oh, my God, 1,000 people.
0:13:32 Oh, my God, 10,000 people.
0:13:34 And there’s like, oh, my God, like,
0:13:36 Football Club Barcelona, like number one soccer club just
0:13:37 used us.
0:13:37 What?
0:13:40 And why isn’t like, how many followers they have?
0:13:42 Oh, my God, 100 plus million followers on Instagram.
0:13:46 OK, I feel like it was actually hard in the sense
0:13:52 of as a founder, you’re like, I’ve put multi amounts of years
0:13:54 into like many things.
0:13:56 And then the thing that we literally are like,
0:13:59 it’s not important gives you all the success.
0:14:01 So it’s a moment of reflection.
0:14:07 You’re like, sometimes like the world throws truths at you.
0:14:11 But those years of work were not all for nothing.
0:14:14 And I don’t think like the years that we’ve been working on
0:14:14 was like a waste.
0:14:15 No, it actually is.
0:14:18 Oh, that’s how you learn on the technical level, how it works.
0:14:20 I mean, because it was so much failure,
0:14:23 we learn about, OK, how do you communicate with your co-founder?
0:14:26 I think that’s something important to note about those times
0:14:30 is that we were very, very aware that this was a trend
0:14:35 and that this was not the end product that we were building.
0:14:38 It was like almost like a marketing engine
0:14:41 that we were using to get better branding and to get known.
0:14:43 They were finding us because of one reason,
0:14:45 but they were staying because of another one,
0:14:47 which was like this other product that we were working on.
0:14:51 Even when we knew that, I think that the core learning
0:14:56 that we got from this experience is that the AI field
0:14:59 changes constantly, like every month or every two months.
0:15:03 There are new breakthroughs, new techniques, new ways
0:15:04 of doing things.
0:15:07 And the tool that we were building,
0:15:09 it was like already starting to get too complex
0:15:12 because we were trying to put everything in a single tool.
0:15:14 And I think that what we learned with the experience
0:15:18 of the spiral virality is that there’s a lot of value
0:15:23 on simplifying super niche and simple use cases.
0:15:29 And that was the case again when LCMs were released.
0:15:32 We saw this technology, and at that moment,
0:15:35 we used all the experience that we got from the first virality
0:15:36 to engine the second one.
0:15:39 And the second one, we knew that it was not a trend.
0:15:41 It was something extremely valuable.
0:15:44 We were like finally being able to get that interaction
0:15:46 that we were looking for for almost years, right?
0:15:49 Like we can generate images in real time
0:15:51 and have full control of the colors, the composition,
0:15:52 the shapes, everything.
0:15:55 That was almost like a dream come true.
0:15:58 Victor and Diego have now hit virality several times over,
0:16:00 but can you engineer that momentum?
0:16:04 In some cases, it’s all about having a single critical feature
0:16:05 not offered elsewhere.
0:16:07 Mohamed from IDU Grimm.
0:16:12 So basically, it was the version 0.1, as we called it.
0:16:17 And this is back then in September of 2023.
0:16:19 And it was a model that was working.
0:16:26 It wasn’t perfect, but we felt like it’s already good enough
0:16:27 to give it to users.
0:16:30 And it was the first model that could put legible text
0:16:31 into images.
0:16:35 So it kind of went viral because of the unique capability
0:16:40 of the model, somehow the ability to put text into images
0:16:42 felt needed.
0:16:46 But in other cases, it’s about cleverly enabling the masses,
0:16:49 or in this case, the memesters, by drastically reducing
0:16:51 the barrier to participate.
0:16:54 Here’s hung with fake old story.
0:16:55 It went pretty viral, right?
0:16:58 What was that like experiencing to put a product
0:17:01 in the hands of so many users and also see that kind
0:17:02 of spread on its own?
0:17:06 Yeah, it was– we didn’t anticipate that for sure.
0:17:09 In the very beginning, we were thinking most targeting content
0:17:10 creators.
0:17:14 But somehow the meme makers and memesters stayed catch up on it.
0:17:16 And that’s how it got pretty viral.
0:17:19 And also, thanks to some of the templates,
0:17:22 we spent so much time discussing why this is the case.
0:17:25 There was this template, the Joker Lil Yachty coming
0:17:26 on the stage.
0:17:28 And there’s a Joker character that
0:17:29 we placed down on the video.
0:17:33 And we’ve seen millions of different characters
0:17:35 just remixing the same moment.
0:17:38 And we realized that the main reason was used to use.
0:17:42 It’s so easy to– basically, you can upload one image.
0:17:44 And then one click, choose that template.
0:17:47 And then in just a matter of seconds,
0:17:51 you’ll have yourself basically in that same moment.
0:17:54 Maybe one other aspect of the virality is, as you said,
0:17:56 the meme makers got ahold of it.
0:17:59 There’s this kind of fun, maybe even silly, aspect to it.
0:18:00 How have you thought about that?
0:18:03 Well, I think that speaks to the entertainment value.
0:18:07 And for anything to have real entertaining value,
0:18:07 it has to work.
0:18:09 It has to work well.
0:18:12 And that actually requires a lot of rigorous in the research
0:18:13 side.
0:18:16 So we are pretty serious about being silly.
0:18:20 And it takes quite a bit of rigorous research to do that.
0:18:22 And the second thing is, you have
0:18:25 to have a tool that provides precise control.
0:18:28 And then because people are getting what they want,
0:18:31 they can have all kind of a variety of fun with it.
0:18:33 You’ve mentioned characters and templates a few times.
0:18:35 What are some of your favorite examples
0:18:37 of those generated on the platform?
0:18:40 One is the Joker/Kamiyanthus state template.
0:18:42 That one is basically the moment we realize,
0:18:45 well, actually, people want to remix.
0:18:48 And there is this virality and memes aspect of it.
0:18:53 And the second one is, there has been one Rakuten advertising
0:18:53 song.
0:18:55 And people are dancing with this.
0:18:57 And we’re also seeing millions of people remixing
0:18:59 that same template.
0:19:02 And this is interesting for us, because you make us realize
0:19:06 that as long as there is this fun elements to it,
0:19:08 people actually don’t mind this content
0:19:10 having a little bit of a brand message.
0:19:11 When you think about applications,
0:19:14 and I know it’s early days, but have there
0:19:16 been any that have surprised you about the ways
0:19:19 that Vigil has been applied every time a founder creates
0:19:20 a product?
0:19:22 They have applications that they envision.
0:19:24 And then the best products are often
0:19:27 people are using them in alternate ways that surprise them.
0:19:30 Yeah, that was exactly the case for us.
0:19:32 In the very beginning, we were mainly thinking
0:19:34 of movie makers, game makers, using–
0:19:37 this might be a quick animation,
0:19:38 pre-visualization tool for them.
0:19:40 It’s actually pretty useful for that.
0:19:43 And we’ve also seen the early users adopting to that.
0:19:46 But then we never anticipated the meme search.
0:19:50 So since that, we’ll be also providing those templates.
0:19:53 So we’ve been keeping track of the latest trendy dance moves,
0:19:55 sports events, et cetera.
0:19:59 And we’ve also seen content creators hopping on to this.
0:20:01 They are actually reaching out to us,
0:20:04 say, can you feature our dance, our song, on your platform?
0:20:07 And then can we collaborate on promoting
0:20:10 some of those that’s been really interesting?
0:20:13 We ask Connor the same question around what he’s learning
0:20:16 by seeing how the masses are using UTO.
0:20:19 The model we originally launched was a model which
0:20:20 generated 32-second clips.
0:20:22 And so to make it kind of a full track,
0:20:24 you would extend that in various directions.
0:20:27 You would add an intro, add maybe a chorus and an outro
0:20:28 and stuff.
0:20:29 And you would build a song like this.
0:20:31 And you would start with these chunks.
0:20:33 And I suppose we’ve actually come to realize quite quickly
0:20:36 that people’s experience with music when they ask for, say,
0:20:38 a song is actually a lot more focused on that.
0:20:41 So they kind of want a song that begins at the beginning,
0:20:42 maybe ends at the end.
0:20:44 It’s maybe– it doesn’t have to be long.
0:20:45 It could be like a short two-minute clip.
0:20:47 But it has a verse and a chorus and a verse.
0:20:48 And there’s a structure to it.
0:20:52 And so I suppose we actually underestimated just how important
0:20:52 that was.
0:20:54 And so that’s something we were making steps to where
0:20:55 it’s rectifying recently.
0:20:59 11 Labs was also no stranger to the surprising and inspiring
0:21:00 user behavior.
0:21:03 Yeah, I think one of the most surprising ones
0:21:05 was people who had lost their voices
0:21:09 and then had used 11 Labs to bring their voices back to life
0:21:11 and then do the thing they loved doing.
0:21:13 So we had Lori Cohen, who was a lawyer.
0:21:15 She lost her voice one morning.
0:21:18 And a friend of hers helped her replicate her voice
0:21:19 with 11 Labs.
0:21:21 And then she was back in the courtroom delivering arguments.
0:21:24 And that, to me, is just such an incredible moment
0:21:26 because you don’t expect that.
0:21:28 And I think our idea was like, hey,
0:21:29 we’re going to give ideas of voice
0:21:31 with our product and our tools.
0:21:33 But this gave someone their own voice back.
0:21:35 And I think that was such an amazing thing to see.
0:21:38 And we saw that again with a climate activist, Bill
0:21:41 Wheel, who was delivering his award speech.
0:21:43 He suffered from ALS, unfortunately,
0:21:45 but, again, was able to replicate his voice
0:21:46 and then deliver that award speech.
0:21:49 So I think those kinds of things are just like–
0:21:51 you’re like, wow, technology being used in a way
0:21:51 we didn’t see it.
0:21:54 And now we want to lean into that and, of course, help others.
0:21:54 Yeah.
0:21:56 Maybe in the opposite sense, have there
0:21:58 been any applications that you’ve actually
0:22:00 built or designed for where you’re like, everyone’s
0:22:02 going to use it for this, obviously.
0:22:04 Or that’s actually not been the case?
0:22:04 It’s interesting.
0:22:07 When we launched dubbing and automated dubbing,
0:22:08 we thought, yeah, this is it.
0:22:10 Like, everyone’s just going to use automated dubbing.
0:22:10 It’s going to be great.
0:22:13 And of course, with dubbing, one of the most important things
0:22:14 is accuracy, right?
0:22:16 And so automated dubbing, we realized,
0:22:19 people still want a lot of creative control on that.
0:22:22 And so we ended up having to build dubbing studio, which
0:22:26 allowed people to go really fine tune that dub and change
0:22:27 a lot of the content.
0:22:29 And then we also introduced 11 Studios,
0:22:31 which was basically creative teams
0:22:34 that help you dub your content with professionals.
0:22:35 We’re really good at that.
0:22:37 And so we realized that actually was
0:22:40 what people needed more of and not just automate everything
0:22:41 and all the things, right?
0:22:43 And then it actually picked up again.
0:22:45 And this is something even when I was working at Palantir,
0:22:47 you learn, which is like the temptation
0:22:49 to try to automate everything or to use intelligence
0:22:50 for everything.
0:22:52 But actually, there’s so much value
0:22:54 in having someone in the middle and still having
0:22:57 that human touch to take it to that final step with something
0:22:59 we learned with dubbing.
0:23:02 And as these companies get all this new data,
0:23:03 it’s not always easy to figure out
0:23:06 who they should be catering to.
0:23:09 So how do you think about what you build and for who?
0:23:11 Your TAM is everyone, in theory.
0:23:14 I think what we acknowledge is that we probably
0:23:17 have different types of users, like distinctly different types
0:23:19 of users, at the very top being, does someone
0:23:23 in a studio who’s making an album at the very top level?
0:23:26 And then at the end of the scale is maybe someone
0:23:28 on their phone who wants, in a minute,
0:23:31 they want just a funny song to send to their friend.
0:23:34 And those are two very different experiences
0:23:38 and somewhat similar to the kind of output
0:23:41 you can get from just an instrument in general.
0:23:42 Someone can have a guitar at home
0:23:44 that they play just to have fun from time to time.
0:23:46 It’s like a totally personal thing.
0:23:47 It’s not anything necessarily serious.
0:23:50 It’s just a way to express yourself with it musically.
0:23:52 And the same way someone can take that same guitar
0:23:55 and a professional can take it into a studio
0:23:57 and make it part of something fantastic,
0:24:00 we like the technology to basically enable
0:24:02 all ends of the oil parts of that spectrum.
0:24:05 Several are unsprisingly using their flywheel of new users
0:24:08 to inform their decisions.
0:24:11 Yeah, we kind of use our user base
0:24:14 and the prompts that they enter into the system
0:24:18 to decide how to evaluate the quality of the model
0:24:20 and what to prioritize.
0:24:23 What’s interesting is our users used ideogram
0:24:25 to tell us what they want.
0:24:27 So they were like, we want image upload,
0:24:30 we want comment, we want more servers.
0:24:34 So I guess the good news is we already have this flywheel
0:24:36 of users coming and using it.
0:24:38 Some are paid, some are free.
0:24:40 And that sets the vision for us.
0:24:43 Hung from Vigol has actually used these new learnings
0:24:46 to expand who they’re building for.
0:24:49 How are you thinking about who you now build for, right?
0:24:51 Are you pivoting or adjusting
0:24:53 to incorporate these new use cases?
0:24:57 So we are broaden our target audience in this sense.
0:25:01 So we are seeing this as eventually we’re going towards
0:25:04 the direction of a new type of AI power content platform.
0:25:06 And the content platform is really important
0:25:08 to have all these creators.
0:25:11 And those are still the content creators, the artists,
0:25:13 the movie makers, the game makers,
0:25:14 the demo game designers.
0:25:18 They are the sources for all those new ideas,
0:25:20 all those new templates.
0:25:23 And then we’re broaden this into content consumers.
0:25:27 Basically Vigol is a new way to consume content.
0:25:29 Before AI, it was mainly like,
0:25:33 if I like the moment I will share it, I will like it.
0:25:35 But there’s a deeper engagement
0:25:36 you can have with that moment.
0:25:39 I can basically, I love this moment so much
0:25:42 that I want to put my own avatar in it.
0:25:43 It’s almost like in a parallel universe,
0:25:47 I want to see how this looks really if that moment myself.
0:25:49 So this is a new kind of content consumption.
0:25:53 And that’s actually one of the most important aspect.
0:25:58 The variety actually comes from all those creative ideas.
0:26:00 So for us, it’s all about empowering
0:26:02 those creative community first,
0:26:03 making sure they have what they want.
0:26:04 They have the best tool.
0:26:07 They have early access to new features.
0:26:09 They have almost private channels.
0:26:12 They have almost unlimited access.
0:26:13 The team at Krea, on the other hand,
0:26:16 is more focused than ever on experimentation
0:26:18 and their signal for success.
0:26:22 When your users are better at using your tool than yourself.
0:26:25 How I think about it is that every tool that we launch
0:26:27 follows a similar process.
0:26:30 And I think that it all starts with a hypothesis.
0:26:32 And I think that this initial hypothesis
0:26:33 needs to come from the founder
0:26:35 and needs to come from your own intuition.
0:26:38 But we are wrong a lot of times
0:26:40 in the way how we validate these ideas.
0:26:43 And when we are wrong, it’s through listening to the community,
0:26:45 seeing what they do with the tools.
0:26:47 And I think that a good rule of thumb
0:26:50 or something that I found that is a good north star
0:26:52 to realize when something is good or not,
0:26:56 is when your users are better at using your tool than yourself.
0:26:59 And that has been key to me, because with the real time,
0:27:01 I was seeing things that I was like, how the fuck?
0:27:02 Did they create that?
0:27:04 And same thing with the video tool.
0:27:06 Like with the video tool, I was trying to do a demo,
0:27:08 like trying to showcase cool stuff.
0:27:10 And I was trying things, and I was not getting there.
0:27:12 And I was looking at Twitter at all the things
0:27:14 that our users were creating with our product.
0:27:16 And I couldn’t get to that quality.
0:27:18 I couldn’t get to those results.
0:27:20 So I think that every time that your users are using
0:27:23 your product better than what you are, that’s a good sign.
0:27:26 – Meanwhile, Justin and Maxis of it are charting new ground,
0:27:29 but also figuring out new limits.
0:27:31 – Stable diffusion allowed you to make anything.
0:27:34 And so when we launched, I wanted to make sure
0:27:37 that we could continue to support that community.
0:27:38 But it was so diverse.
0:27:42 And there’s running meme of things you can make
0:27:43 with stable diffusion.
0:27:45 And in the front is like somebody making funny memes.
0:27:48 And then there’s a train coming that’s porn, right?
0:27:50 Sure, people know that you can make all of this stuff.
0:27:53 I mean, that’s the point of this tech, make anything, right?
0:27:55 And so it was important for us to say,
0:27:58 hey, we want to be able to support this tech as it develops.
0:28:00 It means that we need to embrace all of it.
0:28:01 And that’s not easy.
0:28:06 It’s been incredibly difficult to set up policies
0:28:08 that allow the creation of all things
0:28:13 in a way that’s not going to hurt people
0:28:16 and to also do it in a way that makes it so that people
0:28:18 still have the level of control that they need
0:28:22 to prevent the creation of content that can’t be there.
0:28:23 – In the beginning, our policies were very straightforward.
0:28:26 They were kind of just like, look, as long as it’s not illegal
0:28:29 and as long as it’s not just ethically, completely debased,
0:28:31 then we’ll let it on the platform.
0:28:35 And we were okay when we had the small enough user group
0:28:37 with kind of leaving it even like that vague.
0:28:40 We found over time that we’ve had to really kind of specify
0:28:42 ’cause it turns out that there are just like subsections
0:28:44 of the internet that are into just the absolute strangest
0:28:46 things you’ve never heard of at all,
0:28:48 which can be really funny, which can be really cool.
0:28:49 And some of that is really interesting.
0:28:51 And some of it is just, oh my gosh.
0:28:54 And it’s like a balancing act of figuring out, okay,
0:28:56 what are, you almost have to grow as a person.
0:28:59 And we created like a council of moderators around here
0:29:01 to on our platform to really kind of like get together
0:29:04 and look at when these new things pop up and be like,
0:29:05 how do we feel about this?
0:29:06 One of the things that really blew my mind
0:29:08 when we were getting into the whole moderation aspect was,
0:29:10 like, oh, we’ll just do what other platforms do.
0:29:11 We’ll do what Imager does, we’ll do what Reddit does.
0:29:13 We’ll just copy kind of like what they’re doing.
0:29:16 And as we kind of dug into what they do,
0:29:19 is they don’t define any of this.
0:29:20 None of this is defined.
0:29:22 We had to come up with terms of how do you define
0:29:23 what a child is?
0:29:25 How do you define what is photorealistic?
0:29:27 How do you define what is and isn’t all of these terms?
0:29:31 But before, really didn’t have any really set definition.
0:29:32 – Perhaps it shouldn’t be surprising
0:29:34 that there are new moderation challenges
0:29:36 since this industry is so fresh
0:29:39 with new ideas coming from a new breed of creatives.
0:29:41 In fact, we heard about this range
0:29:43 in both prosumers and professionals
0:29:45 from most of the founders we spoke with.
0:29:46 Here’s Victor from Korea.
0:29:50 – The range of creatives is quite wide.
0:29:52 Like the kind of people that use Korea
0:29:56 can come from having 20 years of working
0:29:58 in the creative industry and being like,
0:30:00 I don’t know, three the artist
0:30:02 or people doing graphic design
0:30:05 or even photographers or these kind of people.
0:30:07 But we also find a lot of folks
0:30:10 who don’t have a professional creative background.
0:30:11 For the professional ones,
0:30:14 you can find them doing a lot of prototyping.
0:30:17 Like for example, when they start working on a new project,
0:30:21 they may go to Korea to really quickly brainstorm
0:30:22 some ideas that they have
0:30:24 and they would use the real time tool
0:30:26 that we have for that.
0:30:29 And they can do like a very simple sketch
0:30:30 add a text from
0:30:32 and have something that looks super realistic
0:30:34 and that can either give them ideas
0:30:37 and maybe even serve as a final deliverable
0:30:38 depending on what they’re doing.
0:30:41 And when we’re talking about like a less professional creative,
0:30:43 it’s honestly more about having fun.
0:30:46 And they are using Korea for everything that you can imagine
0:30:51 from imagining new walls to creating paintings
0:30:54 to creating like characters or all sorts of things.
0:30:55 – And as more participate,
0:30:57 these new platforms generate new talent
0:31:02 but also new expectations like expectations in speed.
0:31:06 – And on the meantime, what we’re doing is building community
0:31:09 and bringing to the community what they want now,
0:31:11 just focusing on what can we do now
0:31:13 with the technology that is out there.
0:31:17 We are very deep into AI communities
0:31:20 and every time that there’s something
0:31:22 that we think that is valuable from a creative point of view,
0:31:24 we go ahead and we execute it very, very fast.
0:31:27 So the way how we’re working is almost like a video game company
0:31:30 where instead of video games, we are building tools
0:31:33 and every six months or so, there’s a new tool
0:31:35 because the space just happened to evolve in a way
0:31:38 that every six months, there’s a new technology
0:31:40 that you can use in order to make a new tool.
0:31:42 And that’s gonna keep being like that until we get to these
0:31:44 like real-time multimodal systems
0:31:47 that allow us to do something way more interesting.
0:31:51 – This new wave has also shifted people’s willingness to pay.
0:31:53 Back to Anish.
0:31:54 I think the willingness to pay
0:31:57 and the amount that consumers are willing to pay is really high.
0:31:59 And that’s really interesting because for so long,
0:32:02 we’ve had these sort of patronage models
0:32:03 for how to fund the arts.
0:32:05 And there’s been this belief
0:32:08 that there’s a sort of decreasing interest in paying for art.
0:32:09 And instead, we’re seeing the exact opposite.
0:32:12 People want to pay for art and pay for tools to make art
0:32:13 and pay a lot.
0:32:17 So that’s a really, really exciting development to me.
0:32:18 – And this willingness to pay
0:32:21 is also unlocking new business models.
0:32:22 – People make so many things
0:32:24 because it’s a tool for creating anything.
0:32:27 And to see the things that people can create,
0:32:29 whether that’s assets for a game
0:32:31 or videos of flowers that are dancing,
0:32:32 it’s just endless.
0:32:33 The possibilities are endless.
0:32:35 And it’s inspiring to see how people
0:32:36 are kind of pulling it to do new things.
0:32:38 – That was Justin from Civet,
0:32:40 which is also working on a new way
0:32:43 to reward AI artists for their contributions.
0:32:45 – When we were getting this going
0:32:47 and we were really like realizing this could be a business,
0:32:49 was we interacted with a lot of the people
0:32:49 who are doing this creation.
0:32:51 And it’s a lot of time and it’s a lot of money.
0:32:52 That’s a lot of like technical skill
0:32:54 that goes into making these things well.
0:32:55 And people were doing it,
0:32:56 thousands of people were doing it,
0:32:57 just for the love of the game.
0:32:59 Like they just really enjoyed the clout
0:33:01 and the entertainment factor.
0:33:02 They got their position on the leaderboard.
0:33:05 – The leaderboard, oh my God, the leaderboard.
0:33:07 And it became pretty clear that,
0:33:09 look, this is almost like a whole new creator economy
0:33:09 that can come out of this
0:33:11 because it’s a group of people who are putting effort
0:33:14 and love into something that could very easily
0:33:15 become livelihoods for them
0:33:17 if they had even the smallest way to monetize it
0:33:18 based on the number of eyes they’re getting
0:33:19 and uses they’re getting.
0:33:21 So yeah, a very clear goal
0:33:22 from the very beginning was like,
0:33:24 let’s figure out how we can keep the creators monetized
0:33:26 while maintaining the open source ethos.
0:33:28 – We actually just announced something
0:33:30 that we’re hoping to roll out over the next six weeks.
0:33:31 Let me give you a little bit of history.
0:33:33 So we launched a creators program four months ago
0:33:35 and we opened it to a small cohort
0:33:37 of essentially 50 creators.
0:33:39 We opened applications and took essentially people
0:33:41 that met certain criteria
0:33:43 and have been experimenting with ways
0:33:45 that we can help them monetize their work.
0:33:48 What we’ve landed on for this next generation
0:33:50 that we’re hoping to open up in these next six weeks
0:33:53 is making it so that people can earn for the generation
0:33:55 that people are doing on our site.
0:33:56 So if they make a resource
0:33:59 that’s intended to produce a new character,
0:34:01 like a consistent character that they’ve made
0:34:03 and somebody chooses to use that in the generator,
0:34:05 then they’re gonna get their share of 25%
0:34:07 of what we charge for that generation.
0:34:10 So the aim is to make it so that these people have a way
0:34:14 to get essentially paid for allowing the convenience
0:34:15 of using their resource on our site.
0:34:17 – One of the main things that we saw right away
0:34:19 before we had the time to be able to implement
0:34:21 any real monetization stuff for creators was,
0:34:23 we put in a DM system simply because we knew
0:34:25 that there’s a lot of people who are contacting creators
0:34:26 for work outside of the platform.
0:34:28 And because of that, I mean,
0:34:30 we just get untold number of people contacting us
0:34:31 being like, “Thank you so much as platform.”
0:34:32 Because of that, I was able to get hooked up
0:34:34 with Hugo Boss or Hyundai
0:34:35 or some of these other people
0:34:36 who are suddenly using this technology.
0:34:38 And it’s completely changed my life
0:34:40 before I was making $30,000 a year as a waiter or whatever.
0:34:44 And now I’m making six figures doing this whole new thing
0:34:45 that’s a passion for me.
0:34:46 And I have lost count
0:34:48 on the number of people who’ve contacted me about that.
0:34:49 So it’s really cool to see.
0:34:51 So from a services side, we want to kind of enable that
0:34:53 and make it even easier for people to be able to sell
0:34:55 their services, their expertise,
0:34:57 directly to businesses from the system.
0:34:58 – It’s not alone here.
0:35:02 11Labs is also building a marketplace for voices.
0:35:06 – I know you guys are building kind of a marketplace of sorts.
0:35:08 So people can upload voices
0:35:10 or they can use voices that others have uploaded.
0:35:12 – Yeah, I think it’s a really exciting way
0:35:15 to give folks a way to earn passive income as well.
0:35:16 Maybe you were a voice actor
0:35:18 and you weren’t getting the gigs you wanted,
0:35:20 but now you can put your voice out there
0:35:21 and you might become extremely popular.
0:35:25 And we’ve seen people earn quite well on our platform.
0:35:28 And so the library is just a great way
0:35:30 to one, put your content out there.
0:35:32 And we want to partner with more voice actors, honestly,
0:35:34 to have more expressive voices
0:35:37 and then give people great voices to create content with.
0:35:39 So two-way street.
0:35:41 – But it’s not just the marketplace.
0:35:43 It’s also the interface.
0:35:46 – I think we’ve always had the dream of voice interactions
0:35:47 with all our products.
0:35:49 If you think about Star Trek
0:35:51 and Knight Rider talking to his car kit,
0:35:52 it’s something that’s been a part
0:35:53 of pop culture history forever,
0:35:57 but I don’t think we’ve had the quality and the sound
0:36:00 and for it to feel as natural as it should have been.
0:36:02 And so I think we’re getting to that point
0:36:04 where the interactions between large language models
0:36:08 using voice interfaces is becoming incredibly natural
0:36:10 and feels like talking to a person.
0:36:11 And so I do totally see a future
0:36:14 where a lot of this physical interface
0:36:16 that you’re tapping around with is going to just fade away
0:36:18 and you’re going to be able to ask the questions
0:36:21 you want to ask and have the conversations you want to have.
0:36:24 I know her is the hot topic movie of the AI space,
0:36:25 but I think there was one thing in that movie
0:36:28 that stuck with me more than just the interactions
0:36:29 he was having with her,
0:36:31 which was there was this scene in the movie
0:36:34 where everyone was down looking at their phones
0:36:35 and kind of scrolling.
0:36:37 And there’s this inflection point somewhere later
0:36:39 in the movie where actually everyone’s kind of talking
0:36:41 to something in their ear.
0:36:44 And I think that is a very precious take that they had.
0:36:46 And I think we’re going to see more of that.
0:36:48 It’s just going to be natural conversations
0:36:51 we’ll be having with this AI or any interface.
0:36:53 – Yeah, it reminds me of my husband’s grandmother,
0:36:55 says that the first time she ever heard someone talking
0:36:58 on a phone in the grocery store,
0:37:00 she thought they were talking to themselves.
0:37:02 Because all of these new interactions, right?
0:37:03 You’re just not used to,
0:37:05 or the people who go to prison and come out
0:37:07 and 10 years later, they’re like,
0:37:09 why is everyone looking down?
0:37:11 And then I realize that we have these crazy computers
0:37:11 in our pockets.
0:37:12 – Totally.
0:37:14 The thing that I love about AI in particular
0:37:16 and all these AI creative tools is
0:37:19 the magic is you had an idea
0:37:20 and now you can imagine it, right?
0:37:22 You can imagine the image that you wanted
0:37:24 and that was in your head and the dream you had.
0:37:26 And now we’re saying you can imagine the sound
0:37:27 that you’re probably hearing in your head
0:37:29 that no one else can hear yet.
0:37:31 – But it’s not just a new UI.
0:37:34 Perhaps it’s a new approach to modeling the world itself.
0:37:36 Hang from Viggo.
0:37:38 – One thing I really look forward to is,
0:37:40 like I said, the next generation of the model.
0:37:44 So we are really hoping to extend this character model
0:37:48 to more the rest of the world, like objects and the scenes.
0:37:52 And so I think those are two general passes
0:37:54 towards modeling the real world.
0:37:58 One is more on, we’ve seen this pixel level approach.
0:38:00 So diffusion models are really good at that.
0:38:02 But it has this drawback of,
0:38:04 it’s really hard to manipulate pixels.
0:38:08 And the real world is essentially, is really, is physical.
0:38:11 So pixel is not really a efficient representation for it.
0:38:14 But it has the advantage of you can train with any video
0:38:16 and it generates anything.
0:38:20 And the hope there is, if we scale it up to a certain extent,
0:38:22 a controllability will kind of emerge.
0:38:25 But we’re taking another kind of different path
0:38:28 in that we want to nail down better the first,
0:38:32 making sure it’s just as precise, as controllable,
0:38:35 as a graphics engine, and then we scale up from there.
0:38:39 So I think this, how those two passes evolve
0:38:44 and how actually they can be combined into one immersive experience.
0:38:46 As we close out this episode, it’s hard to understate
0:38:50 just how much these tools are shifting, what it means to be creative.
0:38:54 To both existing artists and to those who never would have called themselves artists before.
0:38:56 Connor from UDO.
0:39:00 The threshold for someone going into a studio and recording something like that
0:39:01 was way too high.
0:39:04 Whereas now, the promise of the technology is that
0:39:06 it brings an order of magnitude or two orders of magnitude,
0:39:09 more people into the creative kind of experience, right?
0:39:13 Like, people can express themselves in this way,
0:39:16 but kind of even more concretely, as moments happen in the world,
0:39:19 different cultural moments, you can attach music to them now.
0:39:23 Because it can be dynamically attached to these things in interesting ways.
0:39:24 And this is super compelling.
0:39:27 This is a kind of a market that didn’t really exist before
0:39:30 just because it wasn’t actually possible to explore this way.
0:39:33 I think as well as that, we’ve been fascinated with how
0:39:37 at the top level, say with the existing artists or existing producers,
0:39:40 how this can basically work as an ideation machine,
0:39:45 like a kind of well of infinite creativity that you can just pull from for ideas.
0:39:47 Maybe you have the beginning of a track, you have a riff, you have a beat.
0:39:50 You want to see where could this go from here?
0:39:52 If I remix this a bit, what are variations on this?
0:39:54 And that’s a super compelling thing to do as well.
0:39:58 Again, because it’s something that before took a lot of time.
0:40:02 And so it just accelerates the creative experience for professionals like that as well.
0:40:05 I have yet to meet an artist who’s actually used the products
0:40:07 that is worried about the products competing with them.
0:40:11 The biggest worry that I hear over and over is that somebody is going to take them away.
0:40:16 Diego from Korea with a great reminder of just how monumental this shift is.
0:40:21 I was a creative myself doing graphic design, photography.
0:40:25 I even tried to make video games in flash, motion graphics in After Effects,
0:40:30 digital sculpture in Zbrush, 3D modeling for architecture visualization.
0:40:35 And I was like, it’s almost like I felt the fear of,
0:40:38 “Hey, what’s the point if this thing can’t do everything,” right?
0:40:40 But I don’t think that’s the case.
0:40:49 What I think is happening is that we’re just giving so much power to creatives
0:40:56 that things that were like a job in a way like now you don’t even think about them.
0:40:58 That’s what technology does, right?
0:41:03 Like one day it is a lifetime work to move from the east coast of the US
0:41:06 to the west coast and people die on the process.
0:41:11 So now you’re like, “Oh, it took me like 20 minutes at the line to get to the airport thing.”
0:41:15 And you don’t even think about the fact that you flew like a great god through the planes.
0:41:17 Instead, you’re just thinking at a higher level.
0:41:23 You’re just, I don’t know, flying between coasts to make like bigger things, right?
0:41:26 So I feel like the same is going to happen.
0:41:31 Suddenly like coloring 3D models through texture things and all these repetitive things
0:41:37 like sketching and whatever like you will save so much time like of your life.
0:41:42 Because of not having to do that, you can focus on having even better and crazier ideas.
0:41:49 So I’m really, really, really excited to see what the creatives are going to be able to do.
0:41:51 All right, that’s all for now.
0:41:55 The demos shared during the day were followed by a gallery party at night
0:41:59 showcasing many of the artist’s work of the broader New York City creative community.
0:42:02 So if you want to get up close and personal with these tools,
0:42:07 head on over to a16z.com/aiart to check out their demos and more.
0:42:10 We’ll leave you with a little sneak peek.
0:42:17 Ladies and gentlemen, I am thrilled to be here at the a16z artist retreat.
0:42:19 Yeah, it gets you pumped.
0:42:20 So pumped.
0:42:21 So pumped.
0:42:24 You can generate whatever you want.
0:42:25 That was amazing.
0:42:28 Yes, so it’s a whole body swap.
0:42:29 Wow, this is so good.
0:42:33 We’re also working on a new type of memes.
0:42:36 Actually, I think that is better if we see it in slow motion.
0:42:37 All right, you should come see this.
0:42:40 This changes from being deterministic to being totally random.
0:42:42 Remix.
0:42:49 If you liked this episode, if you made it this far, help us grow the show.
0:42:52 Share with a friend or if you’re feeling really ambitious,
0:42:58 you can leave us a review at ratethespodcast.com/a16z.
0:43:03 You know, candidly producing a podcast can sometimes feel like you’re just talking into a void.
0:43:07 And so if you did like this episode, if you like any of our episodes,
0:43:08 please let us know.
0:43:18 We’ll see you next time.
0:43:27 [BLANK_AUDIO]
On June 27th, the a16z team headed to New York City for the first-ever AI Artist Retreat at their office. This event brought together the builders behind some of the most popular AI creative tools, along with 16 artists, filmmakers, and designers who are exploring the capabilities of AI in their work.
In this episode, we hear from the innovators pushing the boundaries of AI creativity. Joined by Anish Acharya, General Partner, and Justine Moore, Partner on the Consumer team, we feature insights from:
- Ammaar Reshi – Head of Design, ElevenLabs
- Justin Maier – Cofounder & CEO, Civitai
- Maxfield Hulker – Cofounder & COO, Civitai
- Diego Rodriguez – Cofounder & CTO, Krea
- Victor Perez – Cofounder & CEO, Krea
- Mohammad Norouzi – Cofounder & CEO, Ideogram
- Hang Chu – Cofounder & CEO, Viggle
- Conor Durkan – Cofounder, Udio
These leaders highlight the surprising commonalities between founders and artists, and the interdisciplinary nature of their work. The episode covers the origin stories behind these innovative tools, their viral moments, and their future visions. You’ll also hear about the exciting potential for AI in various creative modalities, including image, video, music, 3D, and speech.
Keep an eye out for more in our series highlighting the founders building groundbreaking foundation models and AI applications for video, audio, photography, animation, and more.
Learn more and see videos on artists leveraging AI at:
Find Ammaar on Twitter: https://x.com/ammaar
Learn more about ElevenLabs: https://elevenlabs.io
Find Justin on Twitter: https://x.com/justmaier
Find Max on LinkedIn: https://www.linkedin.com/in/maxfield-hulker-5222aa230/
Learn more about Civitai: https://civitai.com
Find Diego on Twitter: https://x.com/asciidiego?lang=en
Find Victor on Twitter: https://x.com/viccpoes
Learn more about Krea: https://www.krea.ai/home
Find Mohammed on Twitter: https://x.com/mo_norouzi
Learn more about Ideogram: https://ideogram.ai/t/explore
Find Conor on Twitter: https://x.com/conormdurkan
Learn more about Udio: https://www.udio.com/home
Find Hang on Twitter: https://x.com/chuhang1122
Learn more about Viggle: https://viggle.ai/
Stay Updated:
Let us know what you think: https://ratethispodcast.com/a16z
Find a16z on Twitter: https://twitter.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Subscribe on your favorite podcast app: https://a16z.simplecast.com/
Follow our host: https://twitter.com/stephsmithio
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.