Sharp Tech with Ben Thompson - Meta and an Abundance of AI Opportunities, When AI Integrates Into the Feed, AI for UI and One Emailer's Dismay

Episode Date: October 31, 2024

Talking through Ben's piece on Meta and Abundance, including the past, present and future of Meta's value proposition to e-commerce advertisers, plans to incorporate AI-generated content into news feeds, and questions about augmented reality and the next phase of user interfaces. At the end: An emailer highlights potential downsides of the vision Meta is selling.

Hello and welcome back to another episode of Sharp Tech. I'm Andrew Sharp and on the other line, Ben Thompson. Ben, happy Halloween. How you doing? Well, you know, Michael Jordan had his flu game. I think we've made references to that in the past. I think we had like a travel podcast. Of course.

A jet leg podcast, things on those lines. Today we're breaking new ground. What's that? Today is the typhoon podcast. Wow, mid-Typhoon. Let's see if the internet holds up. Yeah, I think the internet will be fine. The infrastructure is great here, not in any danger.

But I did not sleep well because the wind whistling is very loud and annoying. And it may happen during this podcast. It's a little calm right now. But you know what? We're professionals. We will power through. That's awesome. Well, I hope the typhoon doesn't ruin Halloween over there in Taipei.

The wind whistling, that sounds great. but I've never slept through or attempted to sleep through a typhoon. So I don't know. Yeah. No, the problem is it sort of like seeps through, it like seeps through the windows. And like that creates the noise. So, which is a very sort of annoying high pitch. Like a worrying.

Yeah, okay. That makes sense. Yeah. So whatever. We're going to power through it. A typhoon podcast. That's right. That's right.

Well, as for the show, you wrote about meta this week. week. So that's where we're going to live on most of this episode. We got some news that we may hit at the tail end. You led your meta piece by writing that meta is in fact the most well-placed company to take advantage of generative AI. And then as a spicier bonus, you wrote that, quote, meta has the potential to be the most valuable company in the world. So let's start with a first statement. Meta is in fact the most well-placed company to take advantage of generative AI. In the short term, Correct me if I'm wrong, but I believe that refers to meta's ability to leverage AI in advertising. And we've talked about that a little bit in the past, but can you give people a refresher on meta's AI opportunity in ads and why it's so compelling?

How generative AI is going to help meta strategically going forward. Well, so just to sort of zoom out. I've been thinking about this piece for a while. You know, so at some point I just had to put it down. I was going to wait until Meta's earnings came out. Hopefully they'd be somewhat bad and go down and I could pull my old trick of just writing an article saying investors are wrong. But then I realized I was just being a big wimp. And if I just had to stay with my chest.

Right now, there's stocks sky high. So this is the time to not write that sort of piece. I should wait and then really get sort of like, again, which I put at the beginning. Look, I've collected on this all through the run of strategy. So I'm definitely playing with house money to a certain extent and taking. a riskier bet and maybe I'm going to lose it all and it's just all it's all going to be a it's not going to work out and I will have you know I wrote meta myths at the the the Nadea whatever you want to call it. It's definitely not a stock price. Yes. Yeah. I mean that was hilarious. I still get like

thank you notes for people. I'm like you're welcome. I've been writing this for 12 years. I mean like it was actually a very easy piece to write. This might be the opposite where I get hate mail because people buy in and then it sort of goes downhill. So definitely running a little bit of a risk there. But it's just a sentiment that has felt very compelling to me. Maybe there's an aspect of Orion that's sort of layered on to this, even though I had been thinking about it a bit before. But like, you know, it just feels like if you zoom out, big picture in a world where we have these incredible content generation capabilities, the company with the best content channels is going to be sort of very well placed to sort of take advantage of that. That at a high

level is sort of the core thing. Yeah. Well, and I will just say reading this piece was entertaining because I thought back to like 18 months ago when people were starting to really fret about some of the AI spend and particularly in meta's case, they were like, well, where does this go? How are you guys going to use AI? And this piece outlines like eight different vectors where AI investment is going to be incredibly helpful for meta moving forward. So it's just a fun rejoinder to where everybody was like a year and a half ago. And yes, it starts with the short term and generative AI in the advertising business. So take me through that aspect. Well, so we talked about, we've talked about this in the past. But basically, you know, as AI, as ads, I should say,

are moving to this fully sort of black box automated experience. And this has been a, long transition over time where, you know, I remember, you know, 14 years ago, you know, one of my managers at Microsoft, like, pulled this stunt or maybe it was, I don't remember a long ago, maybe 13 years, where he basically took out a custom ad for me so that I would make sure, like, and you could like put in enough specificity where it was pretty guaranteed that I would see it, particularly if you'd sort of bit enough. But it was like, you're just doing it basic, sort of like demographic, you know, targeting and focusing in cities or whatever it might, be what was he advertising to you back at Microsoft I don't remember what it was but it was like

you can't the way it worked is they never want to make it so you can like target one person but you can get it down to like a bucket of like 25 people okay and if you spend enough you can like guarantee that like all those people would see it and so I don't I don't even remember what it was but it was very funny at the time just sort of the experience sort of stands out to me but that was you know what most people think about targeting that's I think it's stuck in a lot of people's minds. It's sort of like, I want to target, you know, a white male between the age of like 30 and 35 in the Seattle area who has interest in tech and blah, blah, blah, you know, with, you know, Milwaukee bucks to. Right, whatever it might be. And it's interesting, if you get too specific, they'll stop you because it's like,

look, this is only going to reach 10 people. That's too few people. And so you do have to actually, you had to pull out sort of variables to make sure that the bucket was large enough, you know, for sort of, you know, obvious reasons. Over the last sort of, you know, 15 years, there's been a, a, just a continuous transition away from that. And by and large, it moved away from you as an advertiser, I'd have thought in your head about who your target audience was and like Facebook would help you find it to, you have business objectives that you want to achieve.

You want to sell this sort of item or you want to get subscriptions to this service or you want people to download this sort of app. And over time, it's shifted to. where you buy based on things that you want to happen. And you only, and this is like an evolution of the old, like part of the brilliance of the Google search model was instead of it being display advertising where you pay per thousand impressions, you would only pay per click. And so this was so attractive to an advertiser because once they've clicked, it's in your hands. They're on your website and you have a chance to sort of secure them as a customer and your long, your sort of lifetime value calculation isn't like a sort of affiliate. fee where it's a one-time thing and they don't even know who you are.

It's you can ideally develop something that is worth something in the long run. That's part of what made search advertising so valuable is that customers, Google was thinking about, this was purposeful, Google was thinking about in terms of what, how much money can we get for one quick, but an advertiser had a slightly orthogonal view where it's like, what's the lifetime value, how much, if we, what percentage of these clicks can we convert not just to one-time customers, but I, Ideally, long-time customers. They're on our site.

We can retarget them with ads. If we get their email address, we can send them email marketing. If they buy something and we can sort of develop this sort of capability. And it was more of a balance where these digital platforms came along and they gave new tools to sort of sophisticated marketers. Yes. Now, part of what makes Facebook sort of anti-fragile, a point that I've been making for a few years, is, and I think you see this happen with digital platforms. And you know a digital platform is a successful platform if it's creating companies that are only possible on that platform, particularly in terms of advertising. Facebook has created tons and tons and tons of these sorts of companies.

Amazon has as well. And all these companies that we sort of referenced it in the past sort of source articles in various things in China. And then they figure out, oh, this pops on Amazon. I can sort of get traction here. And it's kind of a tough game like in the Amazon experience because eventually you're going to be paying up through ads and Amazon might launch a direct competitor to you and they're going to get you on warehouse space and all these sorts of things. Right. But you can move on to the next one. And there's people who have made a lot of money sort of serially doing this, building these sort of products and brands on Amazon.

It's a similar thing with Facebook. Like the entire app ecosystem and gaming and all this sort of stuff that we've talked about that was impacted by ATT. they were impacted by ATT because they were made possible by what ATT cut off, which was the sort of Facebook advertising model where you could directly match an ad on Facebook to not just a quick, but a conversion. So Facebook would have an SDK in an app, right? And that would know who you, so your phone has an ID number, the IDFA. Facebook knows, okay, we serve this ad to this number.

Then you go to the app store, which is, you know, Facebook's not a. part of you download the app you go in and but the app knows what your IDFA is so in the back end they can match oh this person saw this ad and now this app is on this person's phone and Facebook could charge for that and charge a lot more than a cost than not just an impression but a quick because it actually like you got further down the you're paying a premium for advertising because you know that it's a huge premium a huge premium right and in the in the day of of IDFA you knew for sure what worked and what didn't. And so you could go to Facebook.

And those are sales that would not have existed if not for Facebook. That's right. That's right. And so it'd be kind of galling because you'd go to Facebook and say, look, if I can get a user to install my app, I've modeled it. And I know that user is worth, say, $10 to me in lifetime value. They'll play the game for a while. They'll see a number of ads. We know they'll churn out after X amount of time. And this is all sort of like basic sort of marketing stuff.

You weren't in business school. I'm running all these sorts of calculations. And so a user is worth $10. And so here's the deal. If that user is worth $10 to you, how much are you willing to pay Facebook to get that user? I mean, in a rational economic model, you're willing to pay $9.99 because that one cent is literally a sense. Yeah. It wouldn't exist.

And it's kind of, I use the word galling in my article because you know for a fact that Facebook is just taking like they're just, extracting all your margin out from you. And yet at the end of the day, you would not have this customer without Facebook. What is specifically galling is that the company that's creating a product that's worth $10 to a customer or whatever, the customer is worth $10 to that company, that's the company that's creating something of social value,

whereas Facebook is just sort of directing people to different products and whatnot. That's not social value? Well, it is, but is it 99% of the value of the product? If you're a customer that got a game that you really enjoyed or, you know, the joke I always have on dithering with Gruber, he always buys belts from Instagram ads. If you got a belt that you really liked, like you would not have known about that without Facebook. Right.

Now, there's a lot about this article that is dystopian. I know we have some questions and comments about this, which we will get to. But one of the reasons I've always been more positive about Facebook advertising than about search advertising is so much of search advertising. is skimming. It's just a straight up tax because you would have been the first result, but because they put six results above it and let all your competitors bid for it, you also have to bid. I'll tell you what, it's a great point. And what is really galling and has been happening to me over and over again over the last like year or so is I will search for

a specific brand name on Amazon and I'll then have to sift through like eight brands that have paid more for advertising at the top of the search results. And it's, It's just like, what the hell are we doing here? I put in the name of the product that I'm looking for. And I guess that's just a new feature of modern life. I've always said I should have written more about Amazon ads on the way up and all sort of things. Because it's such an obvious thing. It was easy to predict.

And yet there's a bit where to me, they're the most distasteful. And it's kind of funny because Amazon ads are like the Apple blessed ad model. Because Apple doesn't interfere with first party. It's all within their app. All the data is self-contained. So from the Apple privacy. view of the world, Amazon ads are what they want to see in the world. And yet to me, I find their social utility negative. They get in the way to your point of what you're trying to do.

It's so blatantly a tax that is leveraging the fact that people go to the Amazon searchbacks to find stuff. How often do I find things via an Amazon ad I wouldn't have found organically because I know exactly what I'm looking for? Almost never. Right. Like I think the Amazon ads suck. And now, there's a bit where the degree to which they suck is precisely why they are so profitable and why I should have written about them and predicted them coming. Because it's such an obvious model that, of course, it was going to be succeed. And I honestly think this is a bit where my personal distaste sort of got in the way. And again, we're going to get to this dystopian bit. I'm going to give you the whole spiel about I'm doing analysis here.

I'm not sort of trying to pass moral judgments. and it's going to be a bit hypocritical because I look at Amazon and I have to admit, yeah, the ad model irks me so much. I probably haven't given it enough phrase over time. Yeah, fair enough. Whereas Facebook, Facebook more than any other platform shows you stuff you never would have encountered. And they do it consistently.

Yeah. And that's true. So you think about that. Sertecri, as I always joke, you know, playing. on a hard mode by having a name people can't pronounce that succeeds through word of mouth marketing, right? I've had the luxury of just because it's information, it's native to the internet. I came up on Twitter at a time when links were currency. If you share good links, you got more followers, right? All that has obviously been demoted is no longer a thing anymore.

And so I feel bad for people coming along. It's good for me competitively. But you come up in that world, I didn't need to think about advertising ever. I've never taken out an ad dollar for shastnery at this point, I probably never will. There's, there's sort of no need to. I mean, I've always felt like I should. I always talk about this. But it's just, I'm just fortunate to be in a business that lends itself to. Basically, I get organic advertising in many respects.

And my, my, my, my, my subscribers pay me and then also do marketing for me. The occasional forwarding. I love you guys. We're not mad at occasional forwarding. Absolutely. No, there's things like in passport, like when you forward an email, from Chichekery, if someone clicks through a link to a paid article, they actually get to see that paid article.

Like that link is tokenized. Now, if they click another paid article from within the website, boom, they hit the paywall. But that's like part of the thinking is like it's architected for that. So because that people share and they sort of do those sorts of things. So it's certainly obviously it's something that I think a lot about, but I haven't had to do sort of Facebook ads. But if you're like selling a product, there's a bit where, okay, you see my belt. And it's like, where did you get that belt?

Oh, you should go to X, Y, Z. But that is, that's a very limited, like the real world has so much friction, right? And to the extent it can spread. A link spreads so much easier and so much further than a belt does. And so in that world, you need a way to reach customers. And Facebook delivers that. And what they've done over time is it's not on the belt maker. to figure out what their customer looks like.

Number one, they're just a startup beltmaker. They don't have the resources or people or whatever to do that. Why can they exist then? Because Facebook's like, yeah, we can see who buys yours. We have, we just collected astronomical amounts of data. We've run machine learning models on this that pulls together, all these things that sort of correlate together. None of us understands it because it's basically impossible to understand.

It's something only a computer. It's all these sparse models and like vectors and all these sorts of things. and this crazy complicated math. Like if you got someone's profile that is used for targeting, it'd be indecipherable. It's just a bunch of numbers, right? It doesn't really mean anything or whatever it might be. But we just know that this sort of person who no normal marketer

whatever in a million years figure out is correlated to this other person who bought the belt. But for some reason, our machine model has sort of figured out this information because that's what machine model. That's what machine warning does. It finds connections that you would never find otherwise. And we show them a belt ad and boom, you have a number. other sale. And you have no idea how you got it. You don't need to worry about where you got it. All you do is, hey, you made some profit and we made some profit and the customer got a new belt.

It's great. Everybody's happy. Yes. So zooming ahead here, Apple severed that particular model about two years ago with app tracking transparency. Believe it or it or not, it's been, it's been, I think it's been four years now. Oh my God. That Apple severed or is it three years. The whole COVID area. Yeah, it was 2022 in my mine, but the specifics. I think 2021, I think they might have announced it 2020. We should look this up. It's funny because they announced it and they didn't roll it out for a while. So it took a while for it to sort of hit the industry.

And but regardless, yeah, that that queer link was severed where you had the number and you could match the number. Right. And this hurt Facebook, a massive amount. Like, you know, billions and billions and billions of revenue that they never sort of materialized. And by the way, a lot of companies that depended on that. the meantime, went out of business. People lost their jobs. Like, the impact was real. When Mark Zuckerberg took out that ad saying this is going to cost small businesses, money and jobs,

it was 100% correct. But the thing was that matching, there's a bit where Apple was doing the work for you by having that number there. So Snap could build a number matching model. An ad network could build a number matching model. Anyone could use this method. Facebook could do it better because they had more and larger sorts of, just a network effect, they had more advertisers, they had more users and all those sort of thing. But if you're on Facebook and ads are getting pretty expensive, it's like,

oh, I'll take some ads out in Snap. I'll sort of do it there. I'll do X, Y, Z. I'll do an ad network. And actually, wow, this, you know, this works pretty well. The cost is much cheaper because I'm not on the,

you know, part of the network effect is having lots of advertisers. It means there's more demand for supply. And meanwhile, Facebook is sort of, you know, Facebook would hit these walls where they would come out with stories, for example, but then stories would saturate. And Facebook has to be careful about how many ads they put because they don't want to churn users because it's annoying.

But then once Facebook stops growing inventory, because the user base is relatively, it's still growing, but it's not growing like it used to. What happens is that means you have a limitation on the supply of inventory, which means prices go up because all the demand is still there. Prices go up. Suddenly people start looking elsewhere. So Apple severs that. Harts Facebook a ton. obliterates everybody else because their shortcut that Apple basically provided is now gone. And Facebook, what you had to do was you had to figure out, they always sort of probabilistically figure out who to target.

All this machine warning and correlation and who might be the sort of thing. The tough part was knowing who converted. They used to know for sure who converted. That was a super strong signal that could go into the model, all these sorts of things. Now they had to also probabilistically figure out who convert. it on the other side. And that was a hard problem to solve because at the end of the day, if you're selling outcomes, if you sell an outcome and it does, and the user's like, oh, I have a $10 lifetime value. I'm going to buy, we'll say $9. And it turns out actually you have a $5

lifetime value. Like, you're going to be in trouble real quick. So you had to really pull back, be super conservative. And Facebook needed to improve that. And so it's never going to be as good as it was. Like, you're never going to have the precision of, you know, where this user is going to be $10 a lifetime value. Now it's maybe more like, it's going to be $9 to $11 a lifetime value. And so there is some value destruction that is still persist in the context of the amount that an advertiser is maybe willing to buy to acquire a customer relative to their lifetime value is never going to be, it's never going to creep up to that line to the extent it did previously. But it's better than countered by the fact.

Competitors, correct? No one else can do this, right? Other than Google. Like Google can actually build out these models. And in this case, Google figured this out also. Facebook figured out first. They were like six months ahead of Google. And they had this huge takeoff in growth where YouTube is where you sort of see this signal for Google.

Because that is, YouTube's the most similar advertising model to Facebook amongst Google's properties. And YouTube cut out like six to nine months later. And they're, you know, they sort of cracked the code as well. No one else is going to crack this code. They don't have the data. They don't have the infrastructure. They don't have the capabilities. Like ATT to an extent from a strategic perspective, again, billions and billions of dollars were not earned in the meantime.

Right. But in the long run, it just basically cemented Google and Facebook dominance in terms of advertising. Deep in the mode. Yeah. Absolutely. For this sort of direct response sort of sort of advertising. Now, other companies are working to refigure this out. I mean, like, like, App Loven is doing really good in the ad network game as far as apps and advertising.

But it's like it's much more sort of contextual based and things on those lines. And they're doing other stuff by being in the middle in the bidding platform. They can get signal about which, which advertisers want to buy which customers. That gives a signal as to what sort of customer that is. And they can start doing correlation between bids and ask and all those sorts of thing and start to build their own sort of internal models. But it's never, it's not nearly as sort of. That works for games. It works for apps.

Does that work for e-commerce? You know, right. E-commerce is really the one where Facebook really sort of dominates. And so the bit about generative AI here is just a very simple sort of observation. There's two points. Point number one is in this world, the thing about AI is AI sort of thrives in the world of increased variables. because where the combinatorial possibilities are massive,

no one could figure that out other than a computer. And in this sort of world, if you introduce more variables, traditionally that's a really bad thing because it's just the computation cost or you get sort of astronomical. AI just sort of like eats variables and paralyzes stuff not for free per se, but to a much greater and powerful extent than traditional computing ever could. And so you add this idea of sort of generative AI. And this isn't going to scale perfectly.

I know we have sort of a pushback email here, which we'll get to. There's going to be lots of challenges and getting this to work effectively, but the idea that you could just try way more creative that can potentially monetize that much better and make your ads that much more efficient and work even better is going, I think, to be a thing. And it's going to be even more of a thing, the more you commit to the Facebook black box. Like big advertisers still don't want to go all the way in because you go to the black box. It's like, I have no idea what's happening.

I know I'm getting Facebook tons of margin, but I'm getting a return. This will help make the black box. So assuming the general of AI stuff is mostly in the black box. It's not just that it makes the black box works better. It makes the black box more attractive. And the black box is where Facebook makes the most margin. And so they're going to get sort of in this, this more attractive from a business perspective approach to advertising.

And I think it's going to be very compelling. That's number one. Number two is one of the challenges Facebook has had is you start out in the original ads on Facebook or text, right? it's in your feed. Then they quickly move to images. Obviously, images are much more compelling for an advertising. It has a visual component. It's in the feed.

That transferred decently to stories. Like stories were images and you would click through and then you'd see sort of another image. But once you get to video and there's video and stories, obviously, but then reels in particular, you're starting to limit your addressable market for advertisers to who has the wherewithal and capability to make video. Right. And so Facebook has tried to address this. they've invested a lot over time in their advertising tools to help you make ads.

But there's still sort of a gating factor. If you're a beltmaker, are you actually capable of making it? That's a thing. Yeah. All the businesses that we're alluding to earlier, they don't have the resources to create like 10 versions of copy and see what works best.

And so there's a bit all these new formats for Facebook actually, uh, it reduces their addressable market of advertisers just because they're more complex. It's harder for them to make ads. And it's not like the feed where anyone could make a Facebook ad. Like my manager on a lark as a joke can make one, right? Well, there's that. And also, I mean, Instagram stories, you just take a photo.

And whatever photo is grabbing people most becomes like the dominant ad campaign. And it's harder to do that in video. So you bring this generative possibility. And where I think it's really compiling is generative video in Facebook just from least a SORA type generative video model. Why don't I mention the article? I actually just totally forgot. Like, like they're they're building these tools where that's, That's the video part is what's really compelling is now any advertiser on Facebook can get video.

And now they go in the black box. What does the black box do? You can go on Facebook and say, I want to target stories or I want to target the feed or I want to target. 35 year old white guys in Redmond. Yeah. Yeah. The black box says, well, yeah, I mean, that's a good point. So we transitioned away from explicit demographic targeting to just find me the customer.

that's happened with placement where you can go to Facebook and say, I want this particular place for my ad to show up. But the black box will say, I want a result and they will figure out where to put it. And this is actually good for Facebook because if they're trying to populate reels, they'll put ads and reels on your behalf and you didn't ask for reels, but that's how they get sort of more liquidity in the reels market

because they're using their black box sort of inventory to do that. Google is for sure, I think, going to do this for when it comes to monetizing these AI overviews. advertisers are going to be skeptical. Are the AI overview advertising really going to work? Is that going to do something? But hey, if you're in Google's Performance Max, their Black Box product, performance max will put ads there.

And you give Google control, Google will optimize not just for you, but also for themselves. And again, people don't like Google optizing for themselves. But as these get so good, they will do it anyway because I think they'll perform that much better. So Facebook could start now if you go in there and you get video for free and then Facebook, we'll figure out where to put it. It's increasing that addressable market of advertisers, which is increasing demand for ads, which is going to drive prices up because you have more demand for the supply that's there. That's good for Facebook. There we go. Well, and speaking of AI generated content, let's move to statement number two, the spicier take. No, no, there's one more.

No, you skipped over one. This sort of text to click messaging bit. So this is, I think, harder for people in the U.S. to Grock, it's actually very obvious to me being in Asia. There is a lot of commerce that happens. There's a lot of commerce that that happens in, even on Facebook here today or in Instagram, where you see someone will post something like the influencers here are like e-commerce retailers. Like that is the business model. And then you go in and you just DM and say, I want to buy this. And then it's like it's all sort of arranged and you maybe you're sent to a link to pay or sort of XYZ. And it's all sort of ad hoc, but it's a remarkably large ecosystem.

And you see it a ton in China. China's more formalized, particularly because you have all the payment through we chat and stuff along those lines. But you do see it in Taiwan and in Thailand and countries like this. And it's definitely a thing people like buying this way. The challenge in doing that in the U.S. is labor costs. Like to actually do this interaction with customers and assure them,

yeah, we have it in stock. Oh, this will work. X, Y, Z, and to go through that transaction, it's just not viable for these products, given the cost structure of labor in the West, particularly in the U.S. Well, guess what? What are you, what would be a good product for a chat sort of experience to consummate a transaction? This is Benioff's abundance take, you know, here we go, a gentic expansion. It's a great, that's a great tie-in. It's exactly what, what Beniof was talking about in terms of, you just get more workers.

Yeah. This is not eliminating jobs. This is creating an entirely new e-commerce channel that we have evidence works very, very well. There's, it's reasonable to expect that it will work well. Now, maybe it turns out cultures are different, XYZ. It won't, you know, the, the, and there are examples of this, right? The do everything app does not exist in the U.S.

Like the plane is the phone and you have lots of apps on the phone. In China, as I've talked about years and years. years ago, the plane is WeChat. And on top of that, there's many apps and all those sorts of things. Just the platform that matters at a different level, that hasn't happened to the U.S. So, you know, Elon Musk talks about, oh, X is going to be the everything platform. No, it's not. It's like, it's just the market did not develop that way. I'm more optimistic about this e-commerce channel becoming a thing. And this is generative AI enabling that. And so this is, it's generative AI enabling more advertising, but not by generating.

ads by making this ad product click to message viable for retailers in a way it wasn't before because it's backed up by generative AI and these sort of chat agents in the sort of like Benioff sort of model. There you go. Well, there's another vector for all the people who wonder where all this spending is going for meta. Statement number two, though, meta has the potential to be the most valuable company in the world.

So now let's move to the medium and long term. and I'm going to read news from the verge. They wrote, if you think avoiding AI generated images as difficult as it is, Facebook and Instagram are now going to put them directly into your feeds. At the MetaConnect event on Wednesday, the company announced that it's testing a new feature that creates AI generated content for you, quote, based on your interests or current trends, end quote, including some that incorporate your face.

When you come across an imagined for you image in your feed, you'll see options to share the image or generate a new picture in real time. So, Ben, can you explain why this is interesting from a business perspective and or why this isn't Horizons Avatar's Part 2? Because it sounds pretty lame to me, but I don't want to hate on it without seeing what it might look like in practice or what it might mean for the business. Well, what about if it says getting the belt, you get the belt on you and you can see what it looks like or or the shirt or the clothes or whatever sort of it might be? I mean, I think this initial start of the reality of humans is there's all a certain extent a bit of narcissism in all of us. We like seeing ourselves. Like the challenge with doing video calls is are you looking at yourself right now? Are you looking at me?

I'm looking at you and you look great. Despite the typhoon. It's good to hear. But like the, you know, people like to look at themselves. I, I, I, yeah. Wow. I am, I'm a professional writer online, which is by definition, narcissistic. So I think people want to hear what I have to say.

There's a, yeah, no, it's like I, my eyes are always bouncing back and forth. Like, I know I should look at you. I know what's going on. And then it's always like, oh, and I'm like, you know, observing like all my soda stream bottles of my shoulder here. Perfectly fair. Look, I'm going to be texting. as we continue to work our way through the rundown here. But continue.

So number one, I can imagine how this is compelling. Number two, again, I heard through the gray find, not from Facebook, just like through other folks that these tests are actually working very well. People do like them. The interesting business bit is there's a bit where this content is actually it breaks Facebook's business model a little bit. because the model is user generating content which they get for free. That's been free for years. Yeah. And so for them to push forward with this,

that is actually evidence. It must be really compelling. It must actually really move the needle in terms of engagement and people trying stuff because there is a cost attached to this content. It has to be generated. Now, the cost for generation continues to drop to the floor. It will continue to do so.

I've talked, this is maybe another thing I should have added in here. There was a crossover point that, was key to the cloud and key to digital advertising when we stopped thinking about compute costs. It's just like we treated it as if it was free. So when you're doing this crazy targeting, that's computationally intensive to target someone. But no one thinks about that cost because the payoff is so clear that you'll do more targeting, not less. It's not like you're doing. And so we'll get to there with genera AI. Yes, it costs money. But at some point, it's going to be basically treated as

free. It's just a cost of doing business. We'll stop thinking in terms of marginal costs. And and think solely in terms of fixed costs, which is just building out the data centers and infrastructure. And it's depreciation that matters, not the sort of per item. Like Facebook right now, they have a cost of revenue, which is running all their servers. They're not thinking about if we serve this user one more ad,

or we serve them an extra piece of content, what's the marginal cost? No, no one thinks about that, right? And that once you adopt the abundance mindset, a lot more stuff becomes possible. And we'll get there with AI if we're not there already. And this is clearly building to that sort of idea. But suddenly you have this possibility of generating content.

And by the way, we just talked about generating ads. At what point it's the same thing, right? Suddenly all content is ads to the extent it's AI. Why not? And you go back to this black box sort of idea. You bid on, I want to sell belts. Suddenly these belts are appearing in generated content that people click on or amuse by because their faces.

it or their friends are in it or whatever it might be. And by the way, you can opt out of this just to be queer, but we'll see how long that opt out last. But again, we'll get to the dystopian bit in a little bit. But like, like in Facebook doesn't have to include that ad. They'll include the belt. They'll include the belt. They'll include the belt maker pays up, right?

That's what I was going to say. You alluded to that possibility, you know, just beyond generating AI content, AI can figure out what is in other content, including authentic photos and videos. Well, not just that. So it could be not just generative content, but real photos, right? You post a video of this. I have my soda stream bottles over my corner. Soda stream in this theoretical world could bid on if a soda stream is identified in a photo,

that should be clickable and it's a link to buy Soda Stream. Listen, Soda stream, you can cut out the middleman with META and just get in touch with Ben Thompson directly. He is a brand evangelist. I'm an influencer. That's right. No, but this is a bit of an assault on the influencer economy to a certain bit, right? Oh, yeah. Instead of paying an influencer to wear your t-shirt or your hat or whatever the heck they do,

I'm not a brand influencer. I just, I buy all this expensive equipment. And yeah, I think I occasionally write posts about what it is. They could bid on so that if it appears, it is, it is a quickable. It's not just like the influencers to say it or put it in the description. It's a clickable because Facebook could do that. Within the black box, would companies then be purchasing the right to have the organic, authentic photos that include their products amplified by meta algorithms? No, no, they pay for a conversion.

And Facebook can figure out how to get that conversion. It could be by showing an ad. It could be by Facebook's like, oh, we're going to leverage content that has this product in it, make it linkable. People will click on that and convert and we've delivered your conversion. So then the influencer really deserves a cut in that. scenario, but yes. Why? Well, because they made that conversion possible by drawing people to their Facebook account or Instagram account. Yeah, well, in the New York Times, I guess deserves a cut when someone posts a link to their article on Facebook that people click through and read. I mean, perhaps.

I don't. The problem is that in this world, like, sorry, like fairness is not part of the equation. No, I'm not saying legally or anything like that. It's just it is unfair in that scenario. Yeah. Well, it's good to be the platform. It's unfair when Apple sort of did this to Facebook before. You could argue or did this to app makers. Like, again, we can add dystopian injustice.

We can add injustice to our discussion later if you want. I'm not crying for the influencers. It's just kind of a remarkable timeline. It's a fair point. It's a fair point. And I think like the analogy to like newspapers being irritated about this is, is a valid. one. This article is not a endorsement and celebration of this. This is a people are underestimating the extent to which AI can make this platform, even though it's more valuable today than it's

ever been, the amount of green space for them to leverage is actually way larger than you appreciate. Okay. Yeah. Well, and I am very curious to see what the AI content looks like, because from the FT adding to the content from friends and creators that users on Facebook and Instagram typically see Zuckerberg said he thought meta would in the future add a whole new category of content which is AI generated or AI summarized content and Facebook and Instagram we're starting to test different things around this. A friend of ours said that sounds awful. Yeah, well, it's going to start out awful.

All this stuff starts out awful. And everyone's going to complain about it and it's going to be moaning and groaning and people are going protested and influencers are going to say, I'm going to YouTube or whatever my views. By the way, YouTube is going to be doing the exact same thing. Just FYI. All of this is going to be awful. Yeah. And then it's just going to happen.

And we're going to look up in 10 years and it's going to be like every pixel on Facebook is monetizable. Do you worry that you're indexing too much on the way it all played out with Instagram stories and Instagram reels? Because that content came from humans. And I think the stumbling block. for me is I've just yet to see a single piece of AI generated content that's compelling. My point is it doesn't matter. All the content could still come from humans and it can all be monetizable

with AI. No, I see that point. It's, I think the, the idea that meta is going to be integrating AI generated content as part of its bounty of content going forward. I think that meta is driven by data. They know exactly what resonates and what doesn't. And if it doesn't, they won't do it. And if it does, they will. and the fact they've announced they're doing it suggests they've been testing and it works very, very well. So my assumption is that it works well. They will know if it works or doesn't work. They'll have it all modeled out what the costs are.

This is like perfectly knowable. Like does it work or does it not? What's not knowable is like does this drive people away from Facebook and the wrong ones because they're so disgusted with it. Like, you know, notification is a great example. My, you know, once there was some period a few years ago where they just went haywire with Facebook notification. and so I just turned them off completely,

even though I thought I had all the control set to get what I wanted. And guess what? I barely ever check Facebook anymore because that was a draw to pull me in. And some mid-level manager had the wrong KPI, which was like driving engagement in XYZ,

pulled that notification's lever and ultimately killed it, at least for me. And Instagram, by the way, has basically done the same thing. I was going to say, I already use Instagram less than I used to. And if it's like 30% AI generated content in five years, I can't imagine how infrequently I would use it.

And it would drive me away. But I mean, I don't want to overindex on my own taste. I think that you don't know your own taste. This is like people saying that algorithmic feeds will drive them away when it's trans-switching an algorithmic feed. Lots of people were Andrew Sharp saying, I don't want an algorithm of fee. If they do that, I'm going to drive away. And Facebook is like, well, we have the data.

And actually, no, you're going to use it more. and more people are going to use it more. And it's pretty overwhelming that that's going to be the case. And guess what? They were right. And like, again, it's not like they're going to do generative content for ideological reasons because we want to have generative content. They will do it because users like it.

Now, I grant the bit that it may be a situation where short-term KPIs can force you make long-term decisions. Again, they blew it with notifications on Facebook and they proceeded to blow it in the exact same way with Instagram a few years later. Like so it's definitely in the realm of possibility. But I think that the reality is, is I would, I would say there's a, you know, maybe we can make sort of an informal bet that in a few years, you'll be fine with it and you won't have even made a conscious choice to be fine with it.

You'll just continually find stuff that's compelling and interesting and it'll keep you using it. And it's not like, oh, I used to have really good content and now I had this blatant AI stuff. there will be a transition period where it doesn't work well and they will churn. But look at how far this has come in just a few years, right? I think there's a bit of the skepticism that is actually underrating how good this stuff's going to get. Yes. Well, we'll get to the dystopian bits in a bit.

Well, we're almost done with the podcast. Maybe we won't because I basically saying it all the way through. In the smartphone era, you wrote, user interfaces started. out being pixel perfect and have gradually evolved into being declarative interfaces that scale to different device sizes. AI, however, will enable generative UI where you are only presented with the appropriate UI to accomplish the specific task at hand. So this was part of the long-term case for meta and its upside. Can you explain more about what you're describing there and how meta's strengths play into that future? Yeah, I'm going to expand on this in another article.

sort of at some point. We always think about new devices, bring sort of new user interfaces. I'm not sure that's right. They bring new platforms for sure. But there's a bit where the, you know, or maybe like the app sort of application layer. So you go back to the computer,

and on the computer was built the internet. What the internet did was created this new application layer that was device agnostic, sort of worked across different things, could be access from anywhere. And that meant when you got an iPhone, even though the first iPhone did not have an app store, it only had what was built in,

but it had Safari. And Safari was a full-featured browser, unlike previous mobile browsers, which meant you had access to the internet. So day one, you had all this sort of stuff that made it sort of useful and sort of compelling. And then you could build up the platform

sort of around it with apps and all that sort of thing. One of the challenges with thinking about AR and VR to an extent is if you kind of go cold turkey, it's like, what do I do on here? This is the Vision Pro problem, right? And so that's why Apple's like, well, you can use all your iPad apps or whatever. And it's like, okay, I can. It's a little bit of a clunky experience.

But I guess, you know, people think, oh, that's a big advantage for the Vision Pro. They have all these sorts of things. Like, you're not going to want to use an iPad app or a phone app on your AR glasses. Let's focus on AR sort of in particular. You can kind of get with VR. Like, I'm in treating the productivity possibility still having multiple monitors with me sort of everywhere. I think that's very cool. But with AR, you're not going to be walking down the street where I have six monitors in front of me and I'm like walking into a phone pole.

That's that's not how it's going to work. And there's a bit where Apple actually has an advantage here because they have the watch. And you think about what was the problem with the watch as an app as an app platform? Especially in the initial. It's like, yeah, and you're like trying to do like an iPhone type app on there and you have everything on there and it's hard to navigate. And Apple did, you know, over time apps used to be pixel perfect to the exact like specifications of iPhone screen. And they changed it to being more sort of declarative. Declarative is like a web page.

You just say it should be here with these sort of margins and XYZ. But it has to be flexible because you never know what size a browser window will be. That's how apps are designed today. It's like you don't know what size the app window is going to be. So it has to sort of handle that gracefully. And so it can to extent it can even go down to a watch sort of size. But what is a watch actually? really great at. Like you get like a notification and you dismiss it or like you get sort of your

alarm comes on, you press a button to snooze or turn it off. All you want is the only thing that you need right then and what you want to do. And what and this, you can imagine a world where watch apps are actually way better when we get to what I'm calling generative UI where it's creating UI specifically for what is happening right now and what you need to do. And it's just like maybe it just shows up on your watch as like one button or maybe two choice. You have to make a choice, right? And instead of having the whole, all the cruft of an app around it, it just pops up.

There's two buttons, pick one. And you can just imagine what a better experience would be where it's fully contextual to what you're doing. It's super convenient and it just makes the interaction much more sort of enjoyable. and where you wouldn't want to pull out your phone. Why would you? You just look down and tap a button. The theory being that eventually AI will be able to sort of analyze your surroundings and then

tailor the UI to whatever it is that it thinks you need. So right now, developers have to create their UIs. In the long run, UIs can be generated on command. This is like we talked about, let's go to the Metaverse example, right? I've talked about what generative AI is going to save the Metaverse. It's been my thesis for a few years, where instead of, of having to create all these assets and create this sort of world and all these rules of how you do it, it will create the scene as you go. There's that demo that's floating around of Doom. We talked about

a few weeks ago where it shows someone playing Doom, but instead of being rendered, it's, you know, or like having a game engine that's drawing the pixels, it's drawing like whole screen by by whole screen by whole screen. It's hilariously inefficient and it's slow and doesn't work very well, but it's a totally different way to manufacture a 3D world, basically doing it screen by screen instead of pixel by pixel. You can imagine that sort of paradigm being applied to user interfaces, where the app can generate a user interface that the developer didn't need to predict ahead of time, oh, we're going to need this user interface.

It actually creates a interface out of nothing because it knows, oh, the user needs to make a choice. I have the capability to generate a choice UI, and it doesn't have to pull from a library, and it can just do what sort of necessary. Again, this is pretty like crazy stuff. This is well into the future. It's not happening sort of tomorrow,

but the tools are coming into place to sort of do this. And that is the AI, that is the UI we need for AR, like not a sort of wall of icons, pick one and go into it. It's that you're in the world and something comes up and you can just act on it. Like imagine a choice comes up and you do one finger or the other finger to make a choice. And it pops up and you can immediately act on that and do it and accomplish it and be done in like less than a second. And it's just super intuitive and natural.

And that happens through our day as you're wearing the glasses. Yeah. And here suddenly it's like how much lower friction could AR be compared to just your phone's with you all the time? If you think about a split second action taken versus pulling out your phone, looking at it, unlocking it, then doing the, it's actually, this is a fair bit. Oh, yeah. Right. And so this is another AI bit. AI is going to make this possible.

And part of what I think Facebook needs to do in these intervening years of, as they work to get the hardware costs down for Orion, is start working on these sorts of interfaces. Now, this is a place where Google and Apple have an advantage because they own the phones. And Apple, particularly with the watch, Google has their watch program as well. They have a place to manifest this. They're also very heavily invested in the current paradigm. They're a bit like Windows taking, like, let's make a mobile phone. Let's put windows on the phone. Like, you can imagine for Apple or Google, we have an app platform.

Let's put apps here. And Facebook has the freedom. They don't have anything to preserve. they can go straight to the correct interface. Yeah, well, and when you were talking there, on one hand, I was thinking, I don't know, there are so many people who are just addicted to their phone. It's going to be difficult to sever that like muscle memory. Like people don't even resent the extra 30 seconds they spend using their phone every day

because people are just at this point on their phone like 16 hours a day, people like you. But then by the same token, once people experience. a once people experience a world with these AR glasses reducing all that friction, they may find it like unbelievably refreshing to not be reaching into their pocket all day long to use their phone. Well, you might still have your phone. But what happens is, so I've talked about the difference with VR and ARR. No, but it reduces the time that you're going back and forth to the phone in theory.

Yeah, what I've talked about there being two tracks in technology. There is the immersive track and the augmented track. The immersive track is video games, movies, TV, VR. where you decide to enter the experience. And in that experience, it's a great experience. And the fidelity can be super high. You sit down and watch a movie. You go to a theater and you get all the surround sound

and all that sort of thing. But it is a sort of conscious choice to enter into that environment and be fully immersed. And because it's a conscious choice and it's sort of a only one thing can win if you're watching a movie or not in a VR experience. Or you're doing one thing at a time.

it's actually a more limited market. It's a valuable market. Gaming's worth a lot. Movies are worth a lot, but it's not as compelling as, say, the phone. The phone is with you everywhere. That's why it's so much more valuable. It augments your life.

It's always in your pocket. What we're talking about here is actually a shift of the phone out of the augmented world to the immersive world, where like PCs used to make, wow, I have a PC I can compute anywhere. That's like augmenting my life. Then we get a phone and the PC is reduced to a destination device. You go to your desk to use your PC.

You're making a choice. It's less valuable than the phone because the phone's sort of with you everywhere. This portends a possibility of the phone was the augmentation device. Now the phone becomes a destination device. And it's still a valuable market, but it's not as good as something that's with you everywhere. And you're always interacting with. That's what I mean is you're just going to be interacting with it far less than you used to. And people may bump up against that in the short term.

And then as they get used to the new paradigm, it may become unthinkable to suddenly be like, It's like a gradual decline. PCs didn't go away right away. Right. They still don't go away. We're both sitting at PCs right now. But it's declined as a share. It's declined a lot in terms of relative share.

But that's because the phone dramatically expanded our times with computers. Yes. It's also declined in absolute terms because that actual jobs get sort of taken away sort of by the new device. One more note before we get to the dystopian question, meta's AI opportunities are so large and so central to the company's future, you wrote, that there's no question that Zuckerberg will spend whatever is necessary to keep pushing Lama forward. Other companies, however, with less obvious use cases or more dependency on third-party

development that may take longer than expected to generate real revenue, may at some point start to question their infrastructure spend and wonder if it might make more sense to simply license Lama. That was just a fun take that I wanted to get on the record as we imagine what the AI landscape will look like over the next three or four years. Do you have anything to add there? It looks a little worse after Google's results where Google Cloud looked amazing and growing a ton and you have to assume that's a fair portion of that is sort of

AI workloads. Maybe like just the reality is AI actually is there's not going to be a crash. Like there's just, it's so compelling that people are going to find use cases. It's going to sort of soak up this, this usage. The big question is these companies are committed to spending so much money and the model costs are going to continue to get so high. At some point, does someone look up and say, can I do this? Like, right. And right now it's like, yes, because the potential payoff is so large, we'll keep doing it. But is there some point where, you know, meta becomes sort of an arms dealer, where it's like, okay, yes, if I could do my own model from scratch, that's best.

But I could also sort of adapt Lama and use Lama doing it. And, you know, we see Lama has a ton of traction in like consulting in like this, where companies do these basic rag implementations. and like Lama, the open sort of open sourceist model was a was a huge success in that regard. It sort of became the standard. And, you know, I don't know. This is purely speculative. It's just like given the upfront capital costs required to make any of this work,

like that would be the rational market outcome except that like five companies are trying to be the model that everybody ultimately licensed. Right. And all these companies have are very profitable and have cash flow to fund this. So it's more a question of investors getting mad about their return on invested capital. It's not like the dot-com era where companies are going into debt to do this, right? So it's a big difference. There's way more runway for that reason. There is a bit where Zuckerberg can just do stuff other people can't.

Like no investor wanted them to spend all this money on the Metaverse stuff, but they did it because he wanted to and he's in control. And that I think is going to apply to the AI spending as well. Now, we'll see, you know, the results did come out. stock is down, not massively amounts, but people are like, you know, where they're spending a lot of money, more cap-fax, like, what's this actually going to be used for? But I don't think it's going to phase, I don't think it's going to phase them at all. Yeah, well, and it puts you right back where you're most comfortable, you know, zagging from the consensus, buying it on the long-term vision. I'm glad the

the results disappointed and the market is, you know, underwhelmed by all of this. Yeah, I mean, it's not quite before and it's down like 20% or 30% or whatever it was, but whatever. I'll take 3%. Meta. I mean, the other thing is a lot of this applies to Google, right? I think Facebook is more compelling in part because of the type of advertising they do, which is top of the funnel. And because the ARVR angle, Google has Android. They have a smartphone platform.

They have the capability to build these sorts of ads. YouTube is like, you know, they talked about they're starting to use AI to analyze YouTube videos so they can do better recommendations for people. If you watch this video, you're more likely this video. Instead of just using the metadata on the video, they can actually. understand the substance of the video. What was unsaid, but obviously pertinent, is they can target ads better if they know the type of videos they are watching. And by the way, they could start selling these exact sort of products, right?

Like you saw something in a YouTube video. There could be a link put there by Google so you could buy that product right there. Like, like, so AI, it's not just the generative aspect. It's the analysis aspect. It can analyze things like videos and pictures and tie it to sort of a. real world product. And so, you know, and then Google, Google is more directly monetizing through the cloud. And again, the cloud results this quarter were pretty spectacular. And, uh, and that's, that's definitely a big sort of upside. And by the way, YouTube subscriptions. Holy crap. Like,

I always praise, you know, sing their praises of why you should get it. I think Google has cut me a check. I mean, it's been very effective for them. Google and Soda Stream pick up the line. That's right. Carthick says, Andrew. and Ben, I just read Ben's article on meta's AI abundance. My God, I don't want it to come true. A digital stream of content that is indistinguishable from ads. I believe Ben when he says meta can do it, but I'm horrified by where we go next. And I'm not questioning Ben's analysis. I just dislike where this is going emotionally, a lot. Maybe the logical next step is buying meta stock, but I cannot bring myself to do it. I've never bought me.

metastock apart from whatever gets bought via index funds because I dislike the TV on steroids aspect of their products. But now they seem well and truly on the path to being the matrix. Is there a silver lining in Ben's analysis or do I need to start prepping for this kind of future? I don't think routine touching grass is going to cut it if this is the force I'm up against. So do you have any silver linings for Carthick? of course touching grass is the solution to this. It's even more the solution. Yeah.

Sort of counterintuitively, I think a lot of this is the solution to actually a lot of the problems we have today. I was going to say, you alluded to it earlier in terms of the targeted type content and just the content that I'm consuming that I, you know, five years ago would never have thought I would be into. I mean, you were sort of speaking on my behalf, assuming that I'm into like Instagram Reels, for instance, which I'm not really. It's never really caught on with me. But there are. Savor that. Don't let it become a thing.

There are consumption habits that I would have been appalled by 10 years ago. And I would say we're already sort of inhabiting the nadir together. And so whatever this may portend, I don't imagine it can be that much worse or even different than where we are today. Yeah, I mean, let's start with the real thing. So take solace part there. The short form video, the short form video is really awful. It is so addicting.

It's like hours can go away. And like it, I think it's very highly problematic. And but there's a, the broader societal issue, people talk about like misinformation and all these sorts of things or wherever might be. And then there's like the engagement and rage baiting and people getting very upset online and all this sort of stuff. I think by and large, and this is maybe more hopeful than anything, but people talk about the digital world is the real world, right?

And like everything sort of moved online. There's an acceleration aspect where I don't like where we are, and this intermingling and moving to a world where it's very tangible that actually everything online is fake. and it's fake in that it might be real, it might be fake, but you don't know, and just a broader sort of cognizance and awareness that you're sort of immersed in this digital dystopia, which is very addicting and fun and enjoyable. I'm hopeful that we draw an ever actually firmer distinction between the real world and

the digital world. And there may be casualties along the way, people that just get sucked into the matrix and sort of can't come out. But I think this bit we're at right now where it's a hybrid and people interpret what's online into their rural world relationships and real world thoughts and community, that's not very good either. There's a bit where I hope we actually, the matrix becomes so deep that it becomes starkly different to the extent people can clearly distinguish between being there and what I'm more worried about. And I've always been most concerned about like lots of surveillance, for example, right? Where the real world's getting ingested into the matrix.

Like, this is why I do find the idea of like cameras on the street and all these pendants people are doing to record all their conversations so I can analyze it. I and frankly, this is, you could say, well, that's kind of what the AR glasses are going to do, right? We already have that where there's cameras. Now you're just inserting stuff as you're walking along. And so this is where maybe I'm just being a little too optimistic. This sort of melding is sort of inevitable. But I'm hopeful that we do start to get a distinction where this augmentation and immersion distinction is very distinct, right?

No one's concerned about, oh, I went to see a movie and all this, and someone got murdered. It's like, yeah, it was a movie. Right. We've actually done this distinction before. We know that movies are not real life versus real life. Right now, that distinction is. kind of fuzzy. And my hope actually is that we end up in a pretty clear delineation.

And we look at all this digital life as entertainment. And everyone is clear about that. Right. Exactly. And again, I'm worried about the addictive properties. Like, again, the short form videos are just like, it's crack. But I'm hopeful that we get to a more delineated sort of concept between sort of, what's real and a touching grass aspect, even if it is augmented versus the, yeah, I'm in like the, the, the, the, the, the, the, the, the, the, the, the, the, the, the matrix right now. And again, I, that may be optimistic, but we're not going backwards. And I don't think where we're at right now is very great. So I'm okay to roll the dice and sort of go forward just because I, people, and people get acclimated. People don't think about right now as being bad. For those of us that

grew up pre-internet and pre-mobile phones. It's pretty crappy right now. And in lots of aspects. And especially the week before an election, you really feel how crappy it can be. Yeah. And you get sort of your headspace. Yeah. Your head space is totally absorbed in something that you have no control over and that

which is not sort of impacted your day to day life right now. That's not healthy. It's not a good way to live. It's important to step back and say like, okay, I'm here in Taiwan. Boy, this typhoon is impacting my life. My kids are here, home from school, of course. That's like the real world. And yes, it sounds cliche, but if I can be more immersed in the real world because I'm going to get that notification and my glasses and I can dismiss it and I don't have to like, right now our augmentation is a mode change.

I'm with my kids. Now my phone's in my hand and I'm not. I'm somewhere else, right? And maybe we do get into this delineation between the Matrix and the real world. Again, I'm totally admitting I'm being optimistic here. And this is sort of a best case outcome. Well, let's be pessimistic. Just if we wanted to go to the other end of the spectrum, one of the thoughts that

occurs to me is we talked about gambling apps maybe six weeks ago and how absence of friction at scale, society might not be able to really handle that. and it might not be healthy for society. And we talked a lot about targeting on this episode and sort of hyper-personalized content to keep people engaged. Like, I wonder whether we get to a point where that sort of hyper-targeting,

if it's successful at keeping the masses engaged, and most people aren't healthy enough to, like, draw a distinction between online and offline and treat it all as entertainment and keep it at arm's length. Do you think, like, 10 years from now, we could be in a situation where there needs to be some sort of limit on the way any of this is deployed.

Yeah, I'm certainly concerned. I think there's probably be some research at some point where there's certain sort of like personality types or whatever that sort of like makes you more susceptible to these sorts of things. I mean, I'm sure you'd see it with a get. Like some people are just like, they know I can't make a single bet because I'm like sort of all in, right? Right.

And I think there's, you're like that with short form videos. It doesn't compel you, right? A lot of people it does, and they're sucked in, and then that's sort of over. We're going to have to figure out how to manage all this as a society. And friction is important, and there's going to be, it's going to have to be bad. There's also a bit where, you know, and this is, this is like truly dystopian, but maybe there's something to it. Like, AI is going to do a lot of work.

It's going to make very productive people and people with initiative drastically more productive and more able to do stuff. And there's a lot of people that it's the new opiate of the masses. It's like, like UBI is, uh, yeah, you, you have sustenance and then you're fully entertained with your, with your headset. And is that horribly dystopian? Yes, it is. Like, this was a business analysis.

It wasn't a commentary on necessarily, again, I'm optimistic we can get to a better place than where we are. But that optimism is born to the. the fact. I think where we are is actually fairly crappy. And by the way, I've spent a decade documenting the business successes of getting where we are. Like, this is just trajectory. I've said this about trajectory again and again. It is analysis. And when I am putting in my personal thoughts and opinions, partly that's what podcasts are for. It's for me to articulate that. Sotrechore is analysis. It doesn't mean I want it or I endorse it, but I'd be doing you a disservice,

like I did my readers with Amazon ads if I get, if that gets in the way of what I'm offering. And I'm just on the tin. That's what Shredchakry does. And that's not to say it's wrong to offer opinion analysis. But that's just not the role that the niche that I fill. And that very much characterizes this article.

There you go. Well, we hear you Carthick and we've got some other emails that we can hit next week. I know. I thought I'm like, oh, this is all good. Well, we got pretty deep into it. I thought we'd spend a short amount of time. there will actually yeah there's i mentioned one pushback on the generative ads we can get to that on monday um you know there's other news this gethub thing super interesting yes we will hit news next week and i appreciate carthick because that was the one thought that occurred to me as i was reading your article like i have had so much fun on the pod for the last six weeks or so because every episode like we're in the first inning of a whole new game and so we're talking about like agentic possibilities and what that unlocks nuclear energy autonomous driving, like, all of that is so exciting to me.

But if we're going to turn around and like the most valuable company on the planet is basically like this giant advertising clearinghouse and they're putting all of us in headsets for the rest of my life, like pretty sure I'm out on that vision entirely, but who can say what the future holds? And we're all going to be experiencing it together here on Sharp Tech. That's the only promise that we can make on this episode. And Ben, you made it through your typhoon episode. So look at you.

Did you hear any of the wind? It really kicked up a few minutes ago. No, but your internet cut out a little bit along the way there. So anyways, we will circle back on Monday. And I look forward to emails, another mailbag Monday, and a distraction from the election. So people can look forward to that. And then have a great weekend. Sharp elbows is coming in the future.

Not next week. Talk to you later.