Big Technology Podcast - Instagram’s Founder On Why All Social Media Looks The Same — With Kevin Systrom

Episode Date: March 22, 2023

Kevin Systrom is the co-founder of Instagram and co-founder of Artifact, a news app that uses AI to determine your preferences and show stories you might be interested in. Systrom joins Big Technology Podcast to discuss the implications of all social media — Facebook, Instagram, TikTok, YouTube, Twitter — starting to look like each other. We speak about how the rise of AI recommendation feeds impacts the future of competition among these apps, creativity upon them, and distribution for people looking to get a message across. Tune in for a fun, in-depth discussion about social media's future in a new era. --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. For weekly updates on the show, sign up for the pod newsletter on LinkedIn: https://www.linkedin.com/newsletters/6901970121829801984/ Questions? Feedback? Write to: [e-mail]

LinkedIn Presents Welcome to Big Technology podcast, a show for cool-headed, nuanced conversation of the tech world and beyond. Kevin Sistram is our guest today. He's the founder of Instagram, and he's currently building a news app called Artifact, which uses AI to determine your preferences, and then it's short. Shostrom and I spoke while we're both in town and Austin when we were speaking at South by Southwest, and I was super eager to hear his perspective on the implications of the fact that all social media, Facebook and Instagram, TikTok, and YouTube, and even Twitter are starting to

look like each other. Sistram should know better than anyone else, what this means for the future of competition among these apps, what it means for creativity upon them, and of course for the future of everyone's ability to find distribution online. I think you're really going to enjoy hearing from Sistram. I know that I walked away from our conversation with a much deeper understanding of the state of play and social media today, and I hope you will too.

My conversation with Kevin Sistram coming up right after this. Hey, Kevin, welcome to the show. Thank you very much. Great to see you here. It's great to see you in person. We don't get to do these in person too often. No. I don't want to get too far ahead.

So let's just start with Artifact. You know, just do it quickly, but like, what is it and what was the inspiration here? It's a hyper-personalized newsfeed. The idea is that you should be able to open up our app and get exactly the news and written content that you want immediately with no pain, no effort. And that sounds relatively easier. You might say, well, I already do that. I open up Twitter and I get all the content I want. And I would challenge you to use artifact for a couple weeks and see if that's actually true.

Typically what people report is they open up artifact and they select some interests. So they'll say architecture, they'll say recipes, they'll say technology. And with about a couple weeks of usage, you start getting served very specific stories that are along your interests. So for instance, I love Japanese modern architecture. I love Mexican cooking. I also particularly like following along with, you know, everything that's happening at meta, not only because I'm a shareholder, but because I used to work there. What's happening with Instagram?

Instagram product launches, anything in the news space. And it figures that out. And my news feed is that. Now, you might say, okay, I use Twitter or I use Facebook for that now. But my experience and the reason why we started the company was because if you opened up any of these social platforms, you very quickly realize what you're being served. You're being served because someone else decided to post about it. And I think that's wrong in a bunch of ways. One, people are incentivized to post when they are excited or angry about something. So it self-selects for content

that is naturally a bit more controversial, not necessarily interesting and not necessarily relevant to you. When it sparks an emotion. Yeah. So what we decided, was we were going to break that chain, and instead of articles or news being served to you because someone else decided to post about it, we'd just find it because it was relevant to you. So the reason why this is possible now is because machine learning took a turn in 2017, 2018, I'll say, a bunch of things were invented, especially around text that allowed you to analyze text in a way that you couldn't be for. And it allowed us to get to a deeper meaning of an article or something written and form a profile on the user that allows us to serve them better.

And in our prototype of this, you know, took about a year and a half to get to where we are. It was really surprising and really fun and really differentiated. So that's what artifact is. It's a hyper-personalized news feed. So I'm using it and I like it so far. And one thing I'm thinking about is I look at artifact and then I look at TikTok and I look at the new spot. and I look at Instagram and I look at YouTube and Twitter to some extent and every single what used to be social media or media or online media, whatever you want to call it,

every platform like this looks a lot like TikTok. I wonder what you think about this homogeneity that we're starting to see because like now you're part of it, right? And of course it's a different form of content than TikTok. But one of my concerns is that we start to lose the creative parts of the internet when everything starts to use AI to serve you content that you might be interested in. and there are all these four U feeds. What do you think about that?

Well, I challenge the premise, which is that everything's converging, or rather that that's a bad thing. So remember when Facebook launched the news feed, when was that, like 2000, I don't know, five, six, something like that. You're in a better position than no. No, I don't really, I just remember the protests down in. Yeah, I remember the protests.

I was part of those. I was in that dorm room. I mean, I wasn't like holding a sign in front of Facebook headquarters. Yeah. I was in one of those, like, you know, 30,000 strong. the newsfeed, which back then in social media days seemed like a lot of people. But remember that a news feed was a fairly new thing for products back then. Like no one had a feed, not really.

We all take it for granted now, but you then saw LinkedIn, you then saw Twitter, you saw all these services adopt effectively a feed. I mean, Instagram, Twitter, et cetera, it wouldn't exist if the feed weren't. way to serve information at the time. And it just so turns out you could use that to serve people very efficiently. Now, what I would argue is at that time, the best way to figure out what was relevant to you was to let you choose. So you would follow, you would friend people, and your social network became your filter

for relevant content. If I follow you, you post something, I see it. These networks grew to be slightly too large for that to be manageable. I'll give an example at Instagram, if we simply just sorted it chronologically, sorted your feed chronologically, the people that posted the most would get the most attention. And that tended not to be the best for people. Like people would self-report seen too much of one person, where they would self-report missing other people. And it's pretty stark like you even look at, like now Twitter has the for you and the following feed side by side.

For you, I'm getting actually interesting tweets. Yeah. Following, I'm getting like a thousand tweets from Bloomberg. No, it's a mess. Sorry, Bloomberg, but enough with the tweeting. Yeah. So there's a volume issue, there's a quality issue, there's all sorts of it. My point was more, the feed was a very good way of serving content when the follow and friend graph were fairly sparse.

When people hadn't figured out how to game getting followers and distribution. But in general, the system is suboptimal if what you do is you do is you do. just simply let people with large followings be the single source of information in your feed, which is what happens when you go chronological. So anyway, someone's smart, figured out we could use machine learning to take that content that everyone's posting, sift through it, and drop anything that's irrelevant, and focus on the things that you'd be most excited about. That worked fairly well.

Facebook became worth many billions of dollars because of it. But that also had a problem in that we kind of forgot that what we were sourced. was stuff that people posted. I'm going to come back to this point time and time again. Because there was enough of that stuff, it was exciting, but it turns out that, like, that's fairly limiting. Why is it that the only articles that I can consume come from a handful of people who decide to post about them?

I mean, there are, you know, many cases, tens of thousands of articles from high-quality publishers published every single day, at least in English, is very hard to find all, like, the good stuff in all of those. So what I was going to say on everything converging to a for-you feed is actually this is just the next version of a feed. This is us figuring out there's no reason why we should just serve you what other people decided to post. We should just go out and find things for you. One of the beautiful things about TikTok is that literally anyone can become famous.

Okay. I have questions about that. I'm going to definitely ask those. Yeah, but anyone can be famous. It is the most democratic distribution system because it's all about the merit of the content, not whether or not that person had built up a huge following beforehand. Right, right. And I think that's really special. And I think what you're seeing is every service deciding, hey, wait, that model works,

the content-based model. Does this content work or not? Rather than did someone influential post about this. By the way, when I'm- And that's important. But I'm hearing you talk, it does seem like we're going to get to some of these questions about the feed, but it does seem like you're directly positioning your new product against Facebook, which is still that old model of what people share.

I mean, it's certainly good for... You're allowed to. That's certainly good for podcast ratings, but like... But that's why I'm asking you about it. I'll push you on this, which is, I think you said yourself, Facebook and Twitter have figured this out. It's why when you log on to Twitter now, you see so much unconnected content. I mean, I had some team ran a test at Instagram the other week.

I don't know. I logged on to Instagram and every fourth image is some celebrity I don't care about. It's like, do you want to follow this celebrity? Do you want to follow this? Like, some partnerships. person decided we had to go get a bunch of followers for celebrities or something. And I think a big part of that is the realization that who you followed initially on a product is not who you want to follow now. Fun story. Okay. I have a friend who works in banking and he's an investor. And he

signed up for the product. He said, okay, I'm going to use this for stocks and investing. And I'm going to I'm going to use this for stock tips, basically. And in reality, it's what this person likes is amazing architecture, great restaurant openings, San Francisco drama and politics. None of the stated preferences match the revealed preferences. And I think what I'm realizing about social media is that there's this huge divide between stated and revealed preferences. And I think Facebook, Twitter, et cetera, have all realized that to be true.

So now as part of your feed experience, you're seeing that more and more. So everyone's figuring this out. So rather than positioning against Facebook, I'd say we're all riding the same wave. It's just who can capture it first. Okay. Good to have you talk that through. Let me ask you, though, because when you talk about the stated preference, right, it wasn't just one stated preference.

It wasn't just follow your friends, right? There were different ways we would build our follow graphs. Facebook was basically friend and family-centric. Instagram, a little bit of friends and family, but also interests and people you want to follow. Twitter interests, right? So, I mean, to move it all toward AI, I wonder, you know, because, you know, are we going to go from having those distinct platforms with their distinct type of content that you follow to all platforms that are machine learning based, starting to recommend the same stuff to people, and in which case, what's the differentiator there? it might. I mean, there's a world.

Like, if you look at, if you look at the history of tech, like social media in the United States, these platforms tend to be single use, like Twitter is for text, Instagram is for images, TikToks for funny videos, YouTube's for videos. You don't see a lot of convergence. And often when you do, it doesn't work, right? Like, everyone tries to be someone else and it only kind of works. Whereas in Asia, I think you see a lot of these super apps, specifically in China, the super apps that do everything, they do banking, they do chat, they do

social media. And there's a very real possibility that using artificial intelligence, some or generally one of these apps or companies will walk away with kind of the everything app. Mostly because I would argue the benefit you get of having a large user base, of having the best, artificial intelligence for recommendations is enormous. It's the same as if, you know, like Instagram is so great because you get on and literally every celebrity and company exists and you can see what they're posting. The network effects exist there, but there's actually a network effect in data as well, which is to say if you have enough data, your recommendations are so much better than the next company that you could

never catch up. And it's possible. I don't, history would not say that. typically happens in the United States, but it's possible. It's interesting that you're even open to the possibility. And one of the things I definitely wanted to ask you was TikTok started this way. Your company is or your former company, Instagram, still consider it your company? Instagram? Yeah, sure. Sure. I mean, they came late to the game. And obviously they have the machine learning and the user base of Facebook, but they're still not fully committed to that TikTok model. It seems like they're going to go there. But if they, if they, give TikTok this large head start, how are they going to compete? I mean, talking about the way that

you talked about data and, you know, having this advantage that compounds, Instagram seems like, you know, listening to you that it might be in some trouble. People forget that TikTok didn't come out of nowhere. Musically existed. I mean, musically, it was fairly popular with a younger crowd in the United States for a while. And when we looked at them, I met with them. I think I was in Shanghai for a board meeting for a different company. And I went into the office and Musically's office. There were probably like 25 kids who didn't look like a day over the age of 18, all in a room, hacking away on computers, building this amazing product where, you know,

people would effectively just post short videos with music in the background, right? Everyone kind of wrote them off. I just, I mean, I remember, like, being told by, we met with Matt Honan, who was the bureau chief at BuzzFeed. he's like looking to musically, I wrote it off. I mean, I feel like I should have been more curious about it. It was easy to discount, but sorry, go ahead. No, you had the right, you were right to write them off.

Because it wasn't working and they had to sell. Now, could we have predicted that one of the world's leaders in AI would spot an opportunity to say, hmm, if we buy this thing and we take all we've learned on all of our other products using machine learning and just layer it on top of Musically, we can create Doyen and TikTok, Doyan being the Chinese version of TikTok. Wow, what a bet, right? And did that work? Absolutely.

So what they were able to do is basically use the initial enormous amount of engagement that had been bootstrapped on musically to then layer machine learning on and then run away with it. The reason I mentioned this is because Reels is pretty good. I mean, a lot of people knock it as kind of like a, you know, there's a wannabe TikTok or whatever, but it's pretty good. And getting better. And it's getting better.

And why is it getting better? More data. You said two point something billion. Like I've lost track of how many users Instagram has like 3.3 billion. Like lots of people using it every single day. If you can just hang on long enough and collect enough data, you can go from like, remember when Apple Maps launched, everyone was like, oh, this is terrible. And like, yeah, it's pretty good now. Like, you just, you refine things and things aren't great overnight, but you can kind of, you can get there.

I mean, Android for the longest time lagged behind iOS and now, like, you talk to people and it's unclear there's an, maybe there's some difference, but like there's, people report being just as happy on both platforms. I think those tend to go away over time as long as the company that's trying to attack has enough data or has enough users. and like when you were sorry go ahead no that's it when you were on instagram did you try to acquire musically uh i did not um mark we were excited about them but there was no like there was no overture that i remember is that a mistake um no because i don't think yeah i mean listen it's easy to you know like what what is it called decisioning when you like judge judge a decision in the past based on information you didn't have at the time like no one knew by dance was going to come in and layer what they had on on top of it and run away with it um but yes i mean like if facebook or instagram

had acquired musically at the time ticot would not exist and that is net i think positive for all these other companies that are now competing with ticot but could have anyone could anyone have predicted that would happen absolutely not i just want to ask you one more question about this quickly The United States is much more cautious now in terms of approving mergers and acquisitions in the U.S. Do you think it's competitive, competitively an issue that they seem to be, the FTC in particular, seems to be interested in, and the Justice Department interested in blocking more mergers of small companies with bigger companies. Does that give an edge to Chinese companies who might have less of a blocker? Possibly. thing for you because you are now a startup entrepreneur but you've also sold a company to

Facebook so yeah but I'm also a startup entrepreneur probably is like the least likely to join a large company at this point right so been there done that yeah yeah had some experiences um a couple of things on this one is I do wish that there were some more clear guidelines stated a priori about what types of transactions would be okay and what aren't it feels a little bit like you're playing roulette right now when you try to acquire a company like i'm sure facebook internally is just like we literally can't acquire companies right i mean they tried to acquire what like some vr gaming thing i can't remember the name of the company um and these things effectively get blocked

what is what are the principles you can use as an operator whether your facebook meadow whatever Twitter to know what you can do and what you can't do because it doesn't seem right that a company just can't acquire other companies. That seems wrong. But on the China part in particular, do you think there's a competitive disadvantage there? I don't know. I'm not sure. If Facebook or Instagram had tried to acquire musically, that probably would have gone through. I don't know. I mean, I'm watching the, you know, the Figma Adobe thing is super interesting to watch. When they announced that, I thought to myself, I said, like, good luck, right?

Everyone acted like it would be fine, and here we are how many months later. We'll see. But I think there's a balance here, which is to say you can't cut off all acquisitions because, honestly, the amount of investment and innovation behind eventually, capital, from venture capital, comes because people believe there can be exits. If you decide that the only exits you can have are going public, there's, like, that doesn't make the system work, right? That takes off, like, a very important part of the feedback in the system. And I think that can be ruinous. Now, everyone losing their deposits in banks. Can that be more ruinous?

Probably. I would assume so. Yeah, exactly. But there are a lot of forces. and headwinds against U.S. startup culture right now that probably don't exist outside and in China. So is there an advantage to being an outside player? Possibly. But at the same time, I don't know. If you're if you're bite-dance and you're trying to do TikTok in the United States, I mean, literally right now it's existential. Can we operate in one of the world's largest ad markets with one of the highest average revenues per user? Should they be allowed to? That's not for me to figure out. Okay, but what do you think, though? It's not. You're going to say it on the show and then it will happen, but I'm curious what you think.

I think it is completely reasonable to use the same logic the Chinese use to not letting American apps operate in China without impunity the other way around, which is to say that, especially by the way, if that app is one of the foremost ways people consume media and information. I worry actually very little about if someone's looking at what funny videos I watch, like very, very little. I don't care where that data is. It's more the ability to manipulate what funny videos you watch. 100%. And I'm not saying that that is happening, but I think the argument typically is Snowden, et cetera, not like, is this actively happening, but is there a door that allows someone to do it if they so choose? Because the context now, everyone's fine, everything's

going well. But in the future, I don't know. Like, is there a world where another world power has a very large app inside of another large, inside of a, you know, instead of a country that they deem competitive and someone could control what you consume, you know, like, throwing an election on Facebook is hard. You got to, you got to set up all these fake accounts. You got to, you got to, like, form groups and then change the topic of the group. You've got to hide. But, like, if you just control what people consume, I see, like, I'm not.

saying this is happening, but don't you think that someone should pay attention to that threat? I think so. Most definitely. Yeah, no, I think that's spot on and I think you're identifying exactly the problem here. 100%. I think it would be illogical to not pay attention to the problem. Now, does that mean that I think that we should shut down TikTok? No. But like, I think there are some real conversations to be had with the leaders of the company to figure out how to make it work. And honestly, TikTok should want to make it work too. Because they're enormously successful in the United States. And so there should be some outcome.

I don't think it's intractable. Yeah. So I want to talk to you now about something that you mentioned, which is that this type of feed, the AI feed is the most democratic possible feed that you can have. I thought about this before we spoke. And I wondered if this is really the case. Here's what I'm thinking, right?

If you're in a follow graph, right? That means average users, I'm just spitballing here. You have the actual data behind it. But average users probably have a better chance of getting their post scene than if they're in the one AI feed where you only have a few videos or articles coming to the top. So I'm kind of curious like what you think the distribution of content is when it comes to this. Like to me it would seem that yes, anyone can go viral, but it's just the content that gets seen instead of this long tail of content that is created by lots of people becomes this highly. the concentrated bit of content and the algorithm just picks from the best. Now, I'm curious what you think about about that and what actually has a more equal distribution. It's funny. You started

asking this question. I was like, I totally disagree. I'm going to fight you on this. And then I was like, actually, I think you have the point. So I think you're both right or wrong. And I will both fight you and agree with you. All right. So here's how I agree with you. It is true that generally speaking, you're going to have the Pareto distribution. You're going to have, you know, 80% of views are going to be on 20% of the content. By the way, that is also true on Instagram. That is also true on Facebook. I would claim that generally speaking, the long tail is dead in social media. It's very, very difficult to get views on unpopular content. Okay. So the counterfactual without AI is, I think, just as bad, which is what I challenge you on, which is like generally

speaking, I'm not sure your Twitter experience is the long tail. Now, I mean, I'm not sure we want the democratic version that I talked about, which is the follow model. Yeah. So here's what I want to say. I, my point in saying it's more democratic is that generally this, I'm going to summarize, I'm going to, I'm going to very overly simplify how TikTok works. Okay. They basically take content without much information and they pre-flight it to a small number. of people and then they basically, it's a very small lab test on that content. They have enough users that they can collect enough of these lab tests to then decide who they want to, if they want to show that content to more people.

In most cases, it's a dud, but in some cases it can be good and they can give it to an extra thousand people and try it a little bit more. And then they collect basically a larger lab experiment on that content. And then they do that again and then again, until they are fairly confident who this content's going to resonate with. I think that is more democratic than gaming your way to more engagement by either buying followers or just being famous or like the other part is I don't love the idea that people with lots of natural distribution meaning they have lots of followers can just decide to tweet about I don't know the efficacy of masks or something and now their word becomes you know the state the record and that to me feels broken. And instead, I'd rather a world where we can engage with content more on this experimental basis.

So, for instance, there's this whole world of what are called bridging algorithms. And the idea is that you put content out there and, you know, let's take sides of a political spectrum. You can find out who's a Democrat and who's a Republican. And you can see what content tends to resonate with both sides rather than just resonate with one side or the other, and you can promote that content. That is much easier to do in an algorithmic world than it is to do in a follow world, because I choose to follow, I don't know, Joe Rogan, or I choose to follow some super liberal.

I'm trying to, like, come up, I don't use Twitter that much personally. So thinking through this, I think the world in which you can use algorithms to dictate what gets distributed is much more powerful and fair, potential. as long as the people in charge of the algorithm have the right intentions, the right objectives, etc. And that's a topic, I think, for the other half. But anyway, that's my long-winded way of saying. I think it is more democratic. And I think that's a good thing. Okay. Instagram founder, Kevin Sistram, was here with us. He's also the co-founder of Artifact. You can find it at Artifact.News. It's available to everyone on the App Store. I just downloaded it.

I'm enjoying it. Maybe you will too. On the other side of this break, we're going to talk a little bit more about content and what actually gets promoted by the AI and then actually we're going to go deeper into the machine learning side of things, which I think you're going to find interesting. Back right after this. Hey, everyone, let me tell you about The Hustle Daily Show, a podcast filled with business, tech news, and original stories to keep you in the loop on what's trending. More than 2 million professionals read The Hustle's daily email for its irreverent and informative takes on business and tech news.

Now, they have a daily podcast called The Hustle Daily Show, where their team of writers break down the biggest business headlines in 15 minutes or less, and explain why you should care about them. So, search for The Hustled Daily Show and your favorite podcast app, like the one you're using right now. And we're back here on Big Technology Podcast with Kevin Sistram, founder of Instagram. He's also got a new app. I might have heard about it, especially if you listen to the first app at the show. It's called Artifact. You can find it Artifact. News. So I'm actually curious now about the type of content you talked about stated preferences and revealed preferences

stated what you want revealed what you say you want revealed what you actually want and i'm wondering what you know it's kind of interesting because like where we like to think that we're all interested in like high-minded news and when i went through like the sign-up flow was like you're interested in stocks and you're in you know you're some stocks and technology and all that but um but then like there was also like dating and romance and I'm like oh yeah I'm definitely clicking that and I'm I'm fairly certain that, like, my feet is going to be all, like, you know, kind of soap opera stuff. And, you know, I also think, like, when Instagram started, you know, especially looking at, like, the way that you framed it when you began, it was like, literally like, here's a way to take your photos and then make them look a little bit better because your camera isn't as great as it could be on your phone.

And it was awesome for that. But as it evolved, you know, a lot of it became about bodies and TikTok was, you know, maybe started with lip singing. but a lot of it became about bodies. And, you know, and I'm wondering where you think people's revealed preferences are going to go with an artifact and whether we're going to end up in a place that, like, again, goes to, like, the, I mean, maybe, you know, it's okay to give people what they want, but like the lurid versus the thoughtful stories that, you know, I don't know, might make us a more informed society. I think it doesn't have to be either or or. Okay. Like, let's get out of news. for a second, just let's go to Netflix or something.

Yes, there are amazing documentaries. And I've heard this, I don't know how true it is, but I've heard this internally that people love adding to their watch later list. What does it call? Is it like the watch list or like, I think it's, I don't really use it that often. But I know there's a plus button. Yeah, people love adding like super heady documentaries to it. And these like thoughtful movies that win all these awards at Sundance.

And then when they log on, would they actually watch? It's like, okay, I'm going to watch reruns of X, Y, or Z. And to me, that's okay. Like, a lot of people often ask, shouldn't we force people to read more intellectual stuff? Shouldn't we make this content higher quality? It's like, okay, sure, you can do that. But at a certain point, how much are we getting in the way of what people just want to do?

And you were talking about bodies. I laugh only because some of my friends have signed up for Artifact. And they do get really good tech news from it. And they do get, like, they'll be like, I don't know, a health care investor. And they get great news about new healthcare startups. But also they really like Daily Mail. And they like the celebrity section. And like maybe they're really interested in following along with Kanye drama or something.

And one of the things I think that Artifact does better than anything else right now is it doesn't judge. Like you can both be a great investor and also be really interested in, I don't know, a specific celebrity or musician. Why make you a better investor? Maybe, right? But the way I look at this is people don't fit the archetypes that we all have in our heads. They fit other archetypes that exist in reality. And I think to try to work against that does. people a disservice in the long run. Now, there are certain lines I don't want to cross

as a company. We try very hard not to carry publishers. When I say carry, distribute in the algorithm, using the algorithm publishers that have a history of bad behavior or misinformation. So we work very hard on drawing the lines somewhere with quality that means what we're rebuilding is responsible because I truly believe as like super powering what you would do as a company anyway. You need to be very, very careful what you put in the machine. And we're seeing some of that with these chat bots. It takes an enormous amount of time and effort to get them to do safe things only. And I see that as a fundamental challenge. But I don't think that we should place a judgment on entertainment.

I think it's okay to be entertained by a wide, like a very wide spectrum of topics and artifact clearly does that. Like if you want funny dating advice alongside your super intellectual, you know, Atlantic article on education, go for it. I think that's great. And the nice thing is machine learning doesn't, it doesn't judge. You can be that person. I think that's special.

Yeah, but it also can have compounding effects where it starts. you know, it clouds out the other stuff and it starts feeding you just the dating stuff. Can I address that? Yeah, yeah. Because I think it's super interesting. You want to talk a little bit about the AI. The first question I get when I tell people the idea, you know, when we pre-launch, I said, this is what we're doing.

Or even, you know, here at South by Southwest and walking around, and people say, oh, you're doing this new thing. But doesn't it have this problem with filter levels? The idea that you only read articles that support your, your point of view, confirmation bias. And that puts you in this loop that basically makes you only see the stuff you want to see. Wouldn't it be healthy to see other stuff, other points of view? And this isn't just politics.

This is advice on parenting or schooling or exercise or nutrition. And I think the answer is yes. So, you know, people who study machine learning, this will be not that interesting. but I found it really interesting that generally speaking, the optimal strategy with any machine learning system is some portion of the time, most of the time, to do what you think the user is going to want. And then you set aside some smaller portion of time to kind of randomize the behavior, because that random exploration turns out allows you to discover things you wouldn't have known otherwise. It's also a very effective way to not what we call

tunnel vision. So if you're reading a ton on a specific, you know, I don't know, issue in San Francisco politics, it's important that we explore potentially other publishers or other points of view, not just the one that you're reading about. And that exploration, I think, turns out to be like all of the value in these systems, because it's easy to serve people what they want. It's a lot harder to discover the things that will be helpful in the margins. And it's something so thrilling when you're like on a TikTok, for instance, and you just kind of die. You randomly get served up that test part of some community that you didn't know existed and you just dive into it. And that's what makes the app.

I think you're right. It makes what makes the app so thrilling. So I just want to ask one follow up about the bodies on Instagram because I feel like you're here and we should talk about it. I mean, the biggest complaint that people have had, I think, about Facebook or Instagram that I've heard is that it harms, you know, teams mental health and puts these unrealistic standards of beauty in front of that. Jonathan Haid has this, you know, new piece out talking about how, like, you can definitively track the decline in teens' mental health with social media. I mean, looking back, I'm curious, A, like what you think about it and B, any regrets or anything you would have done differently. I spend an enormous amount of time internally working on specifically these issues, trying to figure out what resources we could surface within the app when you were searching for a specific topic. Like we had all these early on tags around whether it was self-harm or eating disorders or mental health issues and people would do searches on them to try to find content.

And then people would form these accounts to talk about this type of these types of issues. And the question is, what would you do about it? You know, in some cases we would try to pop up, you know, resources, whether it was suicide hotline or eating disorder information. And we would try to take down self-harm accounts. And then there was, it's interesting. There's always this, there's a lot of information that backs up the idea that if you simply take all of it down, people can't find their communities and then they can't find help. So there's always this balance.

But we worked a lot on it internally. It wasn't like a thing we kind of thought about once a year. It was like an ever-present issue. I wouldn't say we solved it, obviously, because clearly the data supports that social media can be difficult in particularly vulnerable populations. But as we grew, I think maybe the biggest regret I have is that, you know, as founders, we just cared about photography. Like, we just wanted people to take great photos and the commercialization of Instagram is not

something I got excited about over time. It's not something I'm particularly proud of. The idea that you have these beautiful influencers and, and, and people that show, I think, like a non-realistic version of the world and that there's an incentive for them to become this and show this because then they can get famous and get these deals. And the influencer culture is something I've struggled with because it wasn't the vision for Instagram.

The vision for Instagram was that the every person could be a photographer and show the world through their perspective. But what it ended up being because of a lot of reasons is that the power of distribution was funneled into a small handful of beautiful, influential, et cetera, people. Part of why I'm working on what I'm working on now is that I hope it would be very difficult for that to be true in publishing with a machine learning algorithm. Because imagine a world where we opened up to self-publishers. I actually love the idea that we can incentivize a lot of people to become creators in publishing.

where they can publish on topics that wouldn't have been economical to publish on in the past. Maybe niche topics. Like, I don't know if you're into, like, specific type of gardening or Japanese high-fi audio systems or something, something really niche. Can you create a creator ecosystem that unlocks good for the world rather just than maybe just envy? And that's super interesting to me. But, yeah, like, I do think basically there's this race to the bottom on a lot of these social media platforms now. Like, who can get creators and who can get the most influential creators and started on YouTube, basically. They basically popularized the term creator account.

Yeah, yeah. But on YouTube, it feels different because it's a little bit more substantial. Like I study, I don't know, music production. They're these amazing creators. And it feels like I'm going back to school. And that feels very different than, like, travel, influence. or fashion influencers, I think, who dominate the Instagram world. So, listen, I don't want to pick on any particular group.

It's more just that the balance, I think, of content in the long run lean towards this unsustainable vision of what a, you know, a perfect person was or a perfect life. I was with someone last night and they just said, you know, everyone's posting on Instagram back in New York City and I now live in the Bay Area and I feel like they're just all living their best lives. And I'm like, can I just tell you, like, they're not? Yeah.

How many times do people have to tell you? tell you that like your Instagram life is not your actual life. People don't believe it. So listen, not working there now. I'm not sure I have much say in the matter. But we always believed that focusing on people and their personal connections rather than popular influencers or celebrities would be the right direction. You know, and they announced, I don't know, I think it was a year ago that they were just going to focus on creators now. That was the main folks. I just feel like that is, I understand why the game theory leads them to do that, but I wish, and I hope internally there's still, there's still like a firm belief that the people

you know and your family and your friends are the core of what makes Instagram, Instagram. Yeah, it's like a reverse network effect now because the creators are so dominant on the feed. Like I've noticed with my post and maybe my generation is moving off of Instagram, but like stuff that used to be, you know, filled with comments and engagement has now dropped off. I know where that engagement is going. It's to the professionals. Yeah. So it is very interesting.

We have just a few minutes left. One interesting thing about your new company that you've sort of mentioned or hinted at, but I don't think has really been explored too deeply unless I missed it, was that you have ambitions beyond news. You've talked about how news is just a beachhead for you, and what you're really excited about is the machine learning technology. We have just a few minutes left, and I wanted to ask you about that. I'm curious what the beachhead.

Okay, so that's the beachhead. Oh, where's the rest of the land and where are you thinking you might go with this? I want to be clear up front that news is super exciting to me. And I think probably, you know, if you think about taking a bet, you want to have an edge. And I think the news world and social media is oversold, meaning people discount it. Because there have been a bunch of people who've tried it in the past and haven't done well. or rather they've been around a while and they're not TikTok or something. That was also true with photos when Instagram came out.

I remember countless people tell me there's no money in photos, you'll never do anything. And obviously that didn't become true. So you might ask, okay, what's different this time around? The reason why I call it a beachhead is because I feel like I care so deeply about machine learning in the long run. The way that, like, I don't know, Mark Zuckerberg might care about computing when he started, or the internet when he started Facebook back in the day. It's not that we don't care about the beachhead. It's just that, wow, this can be so much more in the long run

if you just see that personalization and machine learning is the thing that's going to turn every industry on its head in the next five years. So to me, when I see written content, I see so much more than news. It's not just what's breaking that day, but it's articles on how to be healthier, or it's information about a musician that you really love or, you know, a critique on a piece of art that you may not have seen in the past.

And that, to me, doesn't feel like news. And that's why I, whenever someone calls us a news app and don't get me wrong, like I use that too. Yeah. I kind of cringe because I'd think it's like calling Instagram a photo app. It's like we ended up being so much more than just pretty photos. And I think if you're going to start one of these companies, I like to remind myself, it's like, okay,

if Apple started and just said we're just going to like sell a computer they wouldn't be what they are today they have to build beautiful build and design beautiful products and delight people and I believe what we're doing is saying we are first and foremost a personalization company and we are going to spend the next five to ten years focused on news but man wouldn't it be cool if you could use exactly this technology and go outside so like shopping's super interesting, right? Or, yeah, let's just take shopping. The number of articles that are served on an artifact today that are full of products that people want to buy, whether it's fashion or electronics, whatever, like you can find ways where this could become something more than

just articles by looking into those articles and saying, oh, actually, there are recommendations on products. That's just an example. I'm not sure we'll go that way. But for the foreseeable future, I am focused on how people consume news. But I am excited about personalization in general. I'm hearing this and I'm like, was this going to be like an enterprise thing where like you get the data and you have the engine and you might like license that type of technology out to others? I wish I could tell you. Everyone asked at the beginning of Instagram, did we know where we were going? It's like you have a map and a compass and kind of points north and you go that direction. But part of the fun of this game is figuring out how you can evolve along the way and not knowing in the

moment. It is interesting. We've talked a lot about, like, TikTok and stuff, but I'm curious if you spent some time researching, I think the app was called Tuytow, right? Tootau. Tootia. Sorry. Yeah. Yeah. And which was effectively what you're trying to do, but it was in a Chinese market. Of course, yeah. It was by dance's first hit. It's the only reason they were able to build TikTok. So did you? I, wow, I'm just curious. Like, now that we talk about it, like, is this kind of following that same map of saying, hey, if we get this right, maybe there's other things we can do? And what did learn when you researched how they were so successful? I mean, they did it, right? They built the newsreader that's based on an AI. So it's possible. And so like all these people who are like,

well, there's not going to be a news app that's going to work. Well, it's ridiculous. It's definitely possible. So yeah, I'm curious if you could tackle those two things. Like, you know, is this sort of a similar bite dance strategy? And what did you learn when you researched that app? I respect bite dance a lot. I think what they've done is amazing. And I think they spotted trends well before they became mainstream. In the U.S., machine learning now is so focused on generative AI, which I think is wonderful and exciting. And a great demo, by the way. And beyond just a great demo, I mean, chat GPD is now like a permanent tab in my browser. And I use it like I use it for. I ask it machine learning questions. I'll say like, oh, we're working on this problem and like, how would we tackle this thing?

And don't get me wrong. 30% of time, it's just wrong. But it's very confident. And that's what's fascinating. It's like, you're like, oh, this is probably the right answer. And you think about it for a while and you're like, wait, this isn't the right answer. And then you go back to it and you're like, it's not the right answer. And then it gives you the right answer and you're like, couldn't you have just given me that the first time? But what's cool about the process is it feels like you have a sparring partner, like a mental sparring partner. And it keeps you on your toe. So I like it for that. My point is more that we focus a lot on generative AI in the United States. I think my experience has been in China, they focus mostly on personalization and mostly on social AI.

All the papers that come out that I've read, at least, are really focused on how do we personalize these enormous user bases? How do we make sure people are seeing the right content? And I think they're just ahead in the personalization game. And I think news apps should work anywhere, not just China. I mean, there are things that make it unique to the United States, particularly. the publishers in the United States and the ecosystem and the history with social media companies for each of those publishers and how they feel about social media. But bottom line is I think that, you know, your job as an investor and entrepreneurs look for

patterns and trends and say, could it be different here? So certainly there's a lot of TOTI has history that inspires what we are going after, but it's a different time. I mean, they were fun. They started a long time ago, and machine learnings, changed dramatically, and the ecosystem has changed dramatically. So it's not as easy to say, oh, that stamp works there, just stamp it over here. But it's certainly exciting and inspiring about it in terms of a market opportunity. And then bottom line, I just think by dance, their strategy of taking what they've built and putting it in different verticals is super exciting. They use text as a beachhead. Yeah, yeah, they did. Yeah. And I mean, they've done this,

they've purchased other companies and tried to layer machine learning on them and to, varying degrees of success, but the, you know, I like to think about tech is kind of like making movies. It's like if your first one's a hit, good on you. And sometimes you have to produce a few more before you get another hit. So the question is just like, do you have the volume of production that allows you to succeed eventually? And we're just trying to start where we see a large opportunity in the United States. Okay. So just last question. Yeah. Machine learning. everybody here or everybody listening, myself, we're all thinking about like a couple of applications.

Image generation, movie generation, text generation, like generative AI, chat GPT stuff, personalization recommendations, engines. When we think about where AI is going to develop now, because it seems like this image stuff and tech stuff came out of nowhere, although I'm sure the researchers like we're aware that this was going on, is there anything else that we should be thinking about, or are those three areas kind of like the key areas to focus

on. I think those are the key areas. Now, in general, I think so many of the resources are pointed at the generative AI that I think you will see the most progress made where the resources are pointed. The productization of these cool demos, like it's cool to say three dogs or like three golden retrievers making sushi and get an image of it and go, wow, that's cool. But the commercialization of that technology, I think, is a different thing. The chat GPT one feels like there's clearly something there that needs to be more than just a chat bot in a window with a gray background. And there are teams now solely focused on this at meta, at Google, and at Microsoft using OpenAI as effectively the head start, that something interesting is going to happen in search clearly. And I think it will surprise us all.

Like we all think this is cool. It's almost like these bank runs. You know, like you hear, you know, you hear 10% or you think there's going to be like a 10% problem and it ends up being transformational. Wipes a bunch of companies out. I think that technological revolution tends to be similar and that you hear little glimmers, you think something's going to be bad. And then it's way worse than it actually is. AI has that property where we all saw cool demos in the last year. I think what you're going to see in the next two years will just like defy expectations.

Now, we're not working on generative AI in the way that these companies are. But I wouldn't focus elsewhere. I would just say, how could it be 10 times better than you actually think it's going to be? And that's my sense. Mark Zuckerberg calls you tomorrow, offers you a billion for artifact. You taking it? Even if I wanted to, it wouldn't happen, right? We talked about that earlier.

Lena Con is on the phone giving you the call. So I'm not going to answer the question because I reject the premise. I don't think it's possible. So let's just talk about something else. Sounds good. Kevin, thank you so much for joining. Thank you. It was fun.

Super fun. And that will do it for us here on Big Technology Podcast. Thank you so much, Kevin Sistram, for meeting me at a recording studio in Austin and recording what I thought was a great conversation. Really appreciate your time. Always great to connect with you. Thank you, Swaycon Studios in Austin. Ken for hosting us and helping us with the audio. Really appreciate that. Thank you,

Nick Gwotany, our editor and audio master, uh, for helping us, uh, produce this and make sure that it's ready for air. Thank you, LinkedIn for having me as part of your podcast network. And thanks to all of you, the listeners, really appreciate you coming back week after week. And I know we have a lot of people who've come in from places like this week in tech and the Drink and Brose podcast and Andrew Yang's podcast forward. If you're just giving big technology a for the first time. Please hit subscribe. We do these twice a week now

of flagship interview like this on Wednesday and then a news recap on Friday and I promise you're going to like what we have in store. Coming up Friday, we have a reporter roundtable and try to bring a few reporters in to talk about the news of the week, which I'm sure will include TikTok's appearance in front of the U.S. Congress

to try to not get banned and all that good stuff and plenty more, of course. And we continue to touch on the financial world and the banking system, something you want to miss. All right, that will do it for us here, and we will see you next time on Big Technology Podcast.