Photo-Illustration: Intelligencer; Photo: Dwarkesh Podcast

Meta, along with a few of its other peers in big tech, is in an enviable place. Like Google and Microsoft, its core business — in Meta’s case, advertising across its popular platforms – is doing extremely well. Also like Google and Microsoft, however, it’s investing breathtaking amounts of money in AI and openly imagining a near-term scenario in which its core business instead revolves around products and consumer habits that haven’t yet come into view.

Meta can afford to make these investments now because it’s raking in cash: it made more than $16 billion of profit last quarter, which puts its plan to spend “$64 billion to $72 billion” this year on capital expenditures, mostly meaning AI, into perspective. As for how it’ll make that money back in the future, the company has been vague, even by the low-clarity standards of the AI race. It’s easy enough to imagine how Microsoft, a company that makes most of its money from productivity software, might sell AI assistants. Meta, on the other hand, has historically made money selling ads on platforms where human beings interact with each other about things they care about as people. Where do chatbots fit into that, and how might they be  worth tens of billions of dollars?

In an interview with Dwarkesh Patel, Zuckerberg finally tried to explain. AI is “almost certainly going to unlock this massive revolution in knowledge work and code,” he said, mentioning that companies like Anthropic and OpenAI were targeting enterprise customers. “I also think it’s going to be kind of the next generation of search and how people get information and do more complex information tasks,” he said, disclosing that most of Meta’s current AI use happens in WhatsApp, where people — mostly non-Americans — use it like ChatGPT.

But the real case use for AI within Meta will arrive on more familiar terrain, Zuckerberg argues, describing human interaction as only the longtime head of the world’s leading attention conglomerate could:

​​I also think it’s going to be fun. I think people are going to use it to be entertained. And, you know, a lot of the internet is like memes and humor, right? And we have this like amazing technology at our fingertips and it is sort of amazing and kind of funny when you think about it, how much of human energy just goes towards entertaining ourselves and pushing culture forward and finding humorous ways to explain cultural phenomena that we observe. I think that that’s almost certainly going to be the case in the future, right?

Delivery aside, Zuckerberg’s claim here is pretty straightforward: For Meta, AI is a way to produce and allow users to engage with content, and a way to extend the ways that people already use its platform. Some of this content will resemble direct human interactions, while other content will resemble media. Zuckerberg says that despite current stigmas, people will become more comfortable interacting with chatbots in personal ways, and that “we will find the vocabulary as a society to be able to articulate why that is valuable.” (This is plausible, but so is a scenario in which the “vocabulary” society finds is that of pathology and addiction.

To monetize, Meta will mostly just do what Meta already does. “For the consumer piece, we definitely want to have a free thing,” Zuckerberg said. “And I’m sure that will end up being ad-supported.” As for what these ad-supported and AI-enhanced feeds might eventually look like Zuckerberg got speculative:

You’ll be scrolling through your feed and there will be content that is basically, I don’t know, maybe it looks like a Reel to start, but then you talk to it or you interact with it, and it talks back or it changes what it’s doing or you can jump into it like a game and interact with it… that’s all going to be AI.

It’s worth backing up and trying to see the world — or at least Meta’s products — as Zuckerberg might. In recent years, driven in part by the popularity of TikTok, social media has shifted away from the social and into media, gradually turning platforms founded on user interactions into more passive channels for consumption. In a recent testimony during the FTC’s antitrust trial against Meta, highlighted by The New Yorker, the company shared that the “percent of time spent viewing content posted by ‘friends’” had declined to just 17% on Facebook and 7% on Instagram, and that the company’s platforms served “the general idea of entertainment and learning about the world and discovering what’s going on.” That people aren’t using public social platforms to communicate with one another is something Meta might see as a problem AI could solve; that people are instead using social media to consume lots of content is something Meta might see as an opportunity that AI could be used to exploit further. (The company is also approaching AI social media from the other direction; it included a “Discover feed” in a standalone chatbot app launched this week.)

Taken together, this vision is indeed “kind of funny when you think about it.” Zuckerberg imagines media platforms in which other people are augmented or replaced by chatbots trained on their collective data, with all parties involved coming together to engage with content produced by AI tools trained on “cultural phenomena,” in a space arranged according to opaque and automatic logics. It might seem like a strange destination for the company that built its empire on real-name social networking, but it can also be understood, less fantastically, as the culmination of years of optimization and automation in pursuit of content that’s cheaper, more plentiful, better targeted, and more engaging. What is an algorithmic newsfeed, or an infinite scroll of Reels, if not the automation of formerly human curation? What is Facebook AI slop if not the platform’s remaining participants identifying and meeting user demands with automation? With fewer friends posting on Instagram, and those still present engaging primarily in private consumption, what is a “Like” if not a signal for an automated system, an interaction with a machine that “talks back or changes what it’s doing” by showing you more?

Meta’s AI strategy less resembles a bet on an unrecognizable future than a belief, or a wish, that it can simply be Meta, but more so.(It’s also a bet that Large Language Model-based AI will present more opportunities for entertainment than for work.)  Zuckerberg isn’t just envisioning  a Meta staffed by AI engineers and AI moderators, but platforms that stock themselves with content, allow users to request whatever types of interactions they want, and are even more effective at holding attention, keeping people engaged, and, of course, serving and targeting advertising.