- 15 hours ago
Category
🗞
NewsTranscript
00:00What do you make of this moment with Sora 2 and VO3 and the impact that it is having on all of media?
00:07Yeah, I mean, I'm extremely excited.
00:09I think that AI is going to be our co-pilot and really make media creation easier for everyone.
00:14As AI video gets better, I expect, especially in the creator economy, for creators to start licensing out their likeness, for example,
00:22so they have more time to be able to create content authentically and not have to, like, fly out for a brand deal.
00:28So I'm excited to just see the progress.
00:31Obviously, we need more training data, which is what scale is built off of, to continue to improve these models.
00:36And I do think we are some time away from things really – I don't think that AI is going to replace, like, us anytime soon.
00:46And you said it could help creators not do some of the, you know, the inconvenient work that takes them away from their day-to-day.
00:54So you would advise the influencers you work with at Passes to, sure, license your image and likeness into one of these engines and –
01:02Yeah, I mean, I think that's the future.
01:04I think right now a lot of creators are worried about the misuse of AI.
01:08So, for example, deep fakes and people creating content against their own permission.
01:13But I do think that the world is moving that way, and soon they're going to realize that their time is better spent working on growing their followers, interacting with their fan base, et cetera, over having to spend a week to fly out, like, do a brand deal, et cetera, and realize that, like, it's a way to scale their time and their identity.
01:33We had Adam Masseri, the CEO of Instagram, on the stage today, and my colleague asked him a question about this tweet from Mr. Beast, and I want to put that back up on the screen if we can.
01:46He said, when AI videos are just as good as normal videos, I wonder what that will do to YouTube and how it will impact the millions of creators currently making content for a living.
01:57Scary times. I mean, he – you know, Mr. Beast knows his business, and he seems to suggest that there is some risk of displacement or replacement here.
02:06I think fans really care, and, like, the creator economy is built off of, like, emotional connection, right?
02:12And I don't think that AI creators will be able to have that same emotional connection with the human population.
02:19We've seen AI influencers such as Lowell Michaela come out, and everyone thought it was going to be the next big thing.
02:24So tens of millions of dollars got invested into these AI creators, and still to this day, we only have a handful.
02:31And I think that this is going to just continue to be true.
02:35Where I see AI being used more is, for example, translating all your videos so audiences across the world can understand what you're saying.
02:44Or in a movie, for example, like, everyone's worried about the AI actors, but I actually think AI is better used for set design.
02:50So instead of having to spend hundreds of millions of dollars for some movies, like, building sets and breaking them down, AI is very good at creating backgrounds.
03:00Maybe I'm a sample size of one, but I wrote about this in the screen time issue, how my algorithm on Instagram, on TikTok, is feeding me, I'm seeing dogs doing podcasts and grizzly bears talking to each other.
03:16And, like, with social media in particular, you know, do you feel like, and maybe it's a novelty so people are clicking, but that it is the potential to blot out, you know, some of the stuff that you're building a business on, you know, influencers performing their art?
03:31Yeah, I mean, I think that algorithms are always going to feed what is the most popular, but I think, for example, if someone were to take an influencer and create deepfakes of them without their knowledge, I think there are a lot of companies out there that are using AI to fight this as well.
03:48And every single social media platform will probably have to implement anti-deepfake technology to prevent this from happening.
03:55And this was a question that Adam addressed earlier today, like, what kind of responsibility do you think the major platforms have in, yeah, identifying AI-generated videos?
04:07Right now, it's an option for some creators.
04:10Yes, I used AI.
04:11And in weeding out deepfakes.
04:13Yeah, I would say that transparency is key.
04:16So if someone is uploading an AI-generated video, but they're doing so without any ill intentions, like, they should be marking that, which platforms are doing.
04:25I think the weed out deepfakes, every social platform does have responsibility to partner with some other AI company that is creating, like, an anti-deepfake API and running their images and videos through that, which they're already doing that for the content moderation, for example.
04:38So this would just be another version of content moderation.
04:41And you're an engineer.
04:42By training, how hard of a technical problem is this?
04:45Is it a solved problem, like, to identify AI videos that may not be properly marked or to weed out deepfakes?
04:53Today, it's not that hard.
04:57But in the future, it will be very difficult.
05:00Why?
05:00Why?
05:01I mean, they're just going to get better, right?
05:02They're going to look more and more like the real thing.
05:05Yeah.
05:06Do the platforms have a responsibility to kind of keep up with their own?
05:09Yes, and I don't think that, like, platforms are going to be building this themselves.
05:12Like, I do think there's going to be a vendor, like, a content moderation API that they integrate with, whose entire job is to fight deepfakes.
05:19Okay.
05:20Yeah.
05:20Okay, so not everyone here might be familiar with Passes.
05:25Yeah, describe what the company does, what your goals are.
05:28So we're building infrastructure to help creators monetize their brand.
05:31And this is everything from building tools to help them create revenue streams to providing health insurance for them to figuring out, like, how they can build their long-term wealth.
05:40Because the goal of Passes is that we build, like, unicorn creators.
05:45Okay.
05:45Now, there are some other companies with sort of similar missions.
05:49Patreon, to some extent, perhaps, OnlyFans.
05:53I know that's a bit of a different market.
05:55Cameo, FanFix.
05:57So tell me, yeah, how you think about the competitive space.
06:00I mean, to some extent, the major platforms as well, right?
06:02They want to be partners with major personalities and help them earn.
06:07So what, you know, are you thinking of Passes as a kind of management agency?
06:11Yeah, so I think for Passes, we're really thinking about it as a 360.
06:14Most of these other platforms are only focused on subscription.
06:17But really, we see creators as businesses and entrepreneurs.
06:22So, first of all, like, we have way more tools than really anyone else to help monetize.
06:26But also, we've created a lot of accredited investors because people are making a good amount of money.
06:31We've paid out nine figures to our creators in just three years, which is incredibly exciting.
06:37Less than three years, I think.
06:39So we are, like, I don't think anyone is thinking about it from the fintech perspective as we are.
06:48What are some variables in your thinking that have changed over these past three years?
06:54How has the company evolved today, you know, from where you started it?
07:00What your initial assumptions were?
07:02Yeah, so I think my initial assumptions were that creators would really want AI.
07:10So, for example, I thought, like, okay, we would be able to do AI deepfakes and creators would be very into that because it'd be free money for them.
07:18It turns out that was just not true.
07:20So we completely pivoted from that.
07:22So right now, our focus is on creating tools to help creators automate their, I guess, like, the manual work.
07:30So, for example, like, A-B testing captions, auto-scheduling posts, auto-scheduling mass messaging.
07:35We have AI tools that automatically price pieces of content based off, like, the demographics of their fans, et cetera.
07:40So they don't have to think about that manual tedious work, and they can just focus on creating content because we realize that creators really enjoy creating content, and that's not something they are willing to give up.
07:49And do you have to go and actively recruit creators, or do you find now that they're coming to the platform?
07:55Yeah, so we have a lot of organic growth.
07:57Our creators really do come through word of mouth because they hear that their friend made so-and-so amount of money.
08:02We do actively recruit as well.
08:06But the creators, like, for us, we go about it from, like, a technology standpoint because our top creators aren't necessarily the ones that have tens of millions of followers.
08:13Like, I'd say our sweet spot are creators with 100,000 to a million followers.
08:17So what we really did was scrape the Internet and kind of assign a machine learning algorithm to figure out what their estimated earnings are based off of, like, whether they have subreddits, how often they post on, like, Instagram reels, TikTok, et cetera.
08:32Because, like, we want to detect growth patterns just because not everyone's going to have 100% retention.
08:36So as long as their following count is growing, that's good.
08:38We look to see trends on, like, who is, you know, popping off.
08:42Stuff like that.
08:43Yeah, and Passes is not mutually exclusive with Instagram, TikTok, the other major platforms.
08:49You want your influencers in both places?
08:51Yep, we do because I think that Instagram, TikTok, et cetera, that's really good for breadth and word depth, right?
08:57So we want them to gain as many followers as possible on these other platforms, and then that widens the funnel.
09:03So then a small percentage, you know, come to Passes as superfans.
09:06Okay.
09:07I want to go back to Scale AI for a moment.
09:09So your former colleagues, some of them, maybe most of them, now work at Meta.
09:14What do you make?
09:15I mean, first of all, why do you think Meta did that deal?
09:18And second of all, what do you make of the arms race and accumulation of talent inside the major tech companies when it comes to AI?
09:25Yeah.
09:25So I think Meta did it as a talent play.
09:28They needed a new way to figure out how to recruit talent.
09:30And to them, this is a really great way to scoop not only a scoop of talent, but have someone at the helm that can go out and poach talent from places such as OpenAI, Anthropic, et cetera.
09:43I generally think that competition is good for innovation.
09:47For example, like I welcome competition.
09:49I think that it is harder for startups to catch up because talent has gotten so expensive.
09:56So in order for a startup to really compete, they're just going to have to raise more funding and be incredibly equity generous.
10:02Are we in an AI bubble right now?
10:04I actually don't think we're in an AI bubble.
10:06And the reason why is because it's a lot cheaper to build companies nowadays.
10:09I was talking to one company that scaled from like zero to 90 million in revenue in four months, and they were telling me how Cursor AI is literally coding 99% of their product, which is mind-blowing to me.
10:20But with all these AI tools, pretty much every single employee is a 10X employee or 100X employee right now.
10:24So it's become a lot cheaper to build companies and reach the valuations that VCs are willing to give, which is why I think VCs are being – it's like they're willing to give those crazy valuations.
10:39To embrace the language of Lucas and Matt's podcast earlier today when they talked about if they're buyers or sellers, when it comes to AI stocks right now, are you a buyer or seller?
10:51I'm definitely a buyer.
10:52I have a lot of my – I have a lot of NVIDIA still.
10:57I hope you don't find this crass.
10:59I've heard you identified as the youngest female self-made billionaire surpassing even Taylor Swift.
11:06I would say Forbes identified in that way.
11:08When you're ever surpassing Taylor Swift, you know that's a good thing.
11:12You've done something right.
11:13But just tell us in the time we have remaining, like, what do you feel like you still have to prove?
11:20I think I've always just wanted to show that, like, I can do more than just that, right?
11:25So after scale, I immediately started a venture fund, and a venture fund has done incredibly well, mainly because of Ramp.
11:32We got in on a seed round, and obviously that company is worth tens of billions now, so that's been amazing.
11:36And now I just want to, like, figure out how to build out in consumer, because I think consumer is one of the hardest challenges out there, and it's fun.
11:44I love to really stay now.
11:44Yeah.
11:50I'll see you in the next few I'm just kidding.
11:51I'll see you in the next few I'm just kidding.
11:54I'm looking through the Maraming receives universal results.
11:55Not just with him, but too.
11:57And he can go out to the bottom right now.
11:58Huldo is a whole bigμα, and I love very hard now, so with that기� ancient room, and I love very hard at all.
12:00I love it, and it's really hard to keep that presenter.
12:03Yeah.
12:03So, what you're cute when you're smart here?
Recommended
1:57
|
Up next
1:58
19:43
17:41
7:01
5:25
5:27
5:19
12:24
14:25
9:42
4:58
8:04
14:36
2:18
6:01
6:50
5:22
3:52
18:27
3:25
1:48
4:31
5:41
Be the first to comment