Photo-Illustration: Intelligencer; Photo: Getty Images

The most contentious debate in tech and finance right now is whether or not the AI boom has turned into a bubble, and what the consequences might be if it has. (In addition to your more typical “yes” and “no” arguments on the matter, some venture capitalists are testing out a contrarian take: Sure we are, obviously, but that’s actually a good thing. Drawing on a new global survey of 250,000 adults in 50 countries by analytics firm GWI, though, John Burn-Murdoch at the Financial Times provocatively argues that, amidst all the fervor about AI, another consequential story is unfolding more quietly: “In years to come, we may well look back on September 2025 as the point at which social media jumped the shark and began rapidly accelerating its transition from the place to be seen,” he writes, “to a gaudy backwater of the internet inhabited by those with nothing better to do.”

The numbers tell a story of stagnation and perhaps the beginnings of general decline: Time spent on social media peaked in 2022, led by steep usage decreases among young people. The decline “is not just the unwinding of a bump in screen time during pandemic lockdowns,” the report says, tracing a “smooth curve up and down over the past decade-plus,” and corresponds with higher numbers of users suggesting they open apps reflexively, just “to fill spare time.”

There are some caveats, of course, chief among them the finding that in North America, an outlier, social media growth as slowed but not yet turned around. But the data has backup elsewhere, in surveys and other attempts at measurement. Teens say they’re cutting back. Facebook reported losing users for the first time back in 2022, and Meta has since shifted to metrics like “family daily active people” — a count of how many people interact with any Meta app in a given day — that conceal how a given platform is doing. Snap is officially losing users in North America, X got Elon’d, and third-party tracking firms suggested that TikTok, that last social platform to truly break through, started stalling in 2023. A recent analysis of the latest American National Election Studies by researcher Petter Törnberg found that, between 2020 and 2024:

Overall platform use has declined, with the youngest and oldest Americans increasingly abstaining from social media altogether. Facebook, YouTube, and Twitter/X have lost ground, while TikTok and Reddit have grown modestly, reflecting a more fragmented digital public sphere.

Also, you know, come on: If you’re on any of these apps, are you using them more or less than a few years ago? And for what?

Given how widely used these platforms are, passing peak social media would be more than an economic story, with rippling cultural and political ramifications. But the economic story would be nonetheless consequential and could spill over into the wider economy in interesting and strange ways. These are some of the biggest companies in the world, and that AI “bubble” is substantially financed by cash thrown off by services like Instagram. It’s reasonable to wonder how long social media revenue*,* which has been outpacing user growth for years now, can keep up the pace.

The data might suggest we’re on our way down from the peak, but it doesn’t tell us much about why. Platforms like Facebook were lean, ingenious, and eventually self-perpetuating businesses that, after crossing into the mainstream, seemed destined for long-term entrenchment. What stalled their momentum? And where else are people spending their time?

Everyone’s there already

In 2024, a pair of Pew surveys documented two simultaneous phenomena. One on hand, 45% of teens said they were spending too much time on social media, up from 27% the year before. On the other, nearly half of teens surveyed said “they are online almost constantly,” with 96% using the internet “daily,” sorted across a variety of platforms, nearly all of which have declined in the last two years, with sharper long-term declines for older platforms like Facebook and Twitter, now X. This is an age range for which social media is a set of entrenched options in a saturated market that users can’t help but know about. They have accounts on most if not all of them but only use the ones they want. It’s hard to find growth in a demographic like that if you’re not a brand-new platform, and no new social app has achieved ubiquity since about 2019. Every young person who might become a TikTok user probably already is.

In 2025, this dynamic extends beyond young people. Meta counts more than two billion people as “family daily active people,” which, once you exclude people in countries where its products aren’t allowed, the very young and very old, and people without access to cellphones or the internet, doesn’t leave a whole lot of room to grow. Facebook has been around for 20 years, and Twitter launched in 2006. Instagram is 15 years old and TikTok blew up during Trump’s first term. Generally speaking, if someone was ever going to try them, they have, and they might not be likely to give them another chance.

But also, nobody’s there anymore

Earlier this year, during the FTC’s antitrust trial against Meta, the company shared a wild stat: Just 17% of time spent on Facebook, and 7% on Instagram, was spent viewing content posted by “friends.” The company’s platforms, it said, served “the general idea of entertainment and learning about the world and discovering what’s going on.” A 2023 survey by Morning Consult found a significant segment of social media users were consciously posting less than they used to. At The New Yorker, Kyle Chayka argued that we might be moving toward “Posting Zero,” a situation in which regular users “stop sharing things on social media,” shifting to a purely consumptive mode.

Declines in user posting used to be considered a fatal sign for a social network, and even small shifts in user habits could lead to massive product changes. When Instagram added Stories, it wasn’t just ripping off fast-growing Snapchat, it was dealing with worries that users weren’t posting to their regular feeds as often as they used to, threatening the health of their core product. Giving people a way to make temporary posts in a new format helped the app keep drawing content out of its users while its main feed withered.

More people are worried about posting

There are plenty of reasons why people might choose to post less on social media, and fear is one of them. According to that Morning Consult poll, many social media users are becoming “more selective” about what they post, with the highest rates reported among Gen Z. That the youngest users are most cautious about what they post is both a relatively new phenomenon — be careful what you post on the internet was something parents used to tell their kids, not the other way around — and makes lots of sense: These are people who came online after cancel-culture discourse and in a world where public posting was something influencers did in pursuit of fame or money. This shift has lately shown up in morbid ways, too: If acts of spectacular or newsworthy violence involve people under the age of 35, their revealing “online trails” often fail to materialize, not necessarily because they were hiding anything, but because having a public Instagram profile — or posting a bunch of unguarded stuff online under your real name like it’s 2010 — just isn’t typical anymore.

That you shouldn’t post too freely on social media has become — or returned to — conventional wisdom: If you’re posting publicly, your family might see it, your boss might see it, and, every once in a while, a few million strangers might see it as well. An increasingly dire political situation has added new dimensions to these worries. A core message of the MAGA movement about social media has been, basically, you can’t say anything online anymore, a claim often couched in terms of free speech, but which also merged matters of platform censorship with broader fears of social ostracization. Now that the MAGA movement is in power again, and the social platforms have gotten well and fully in line with the administration, the situation hasn’t so much been resolved as inverted and intensified: It’s the left’s turn to be afraid to post freely, and many on the right would like to make sure of it. One of Törnberg’s most interesting findings in the ANES data — which ends in 2024 — is that “most platforms have moved toward Republican users,” with Twitter’s swing being the most pronounced, leading to a partisan “reconfiguraton” with “Republican users shifting from ideologically homogeneous venues such as TruthSocial into mainstream platforms such as Twitter/X” while “Democrats have retreated toward emerging, smaller networks such as Bluesky, Mastodon, and Threads.”

More generally, an environment of rising mutual suspicion and decreasing social trust — almost certainly caused in part by social media itself — is probably not going to be great for products built on the premise of sharing your every thought with ever-growing and diverse groups of people. Neither is the administration frequently broadcasting its ability and intention to “screen” non-citizens’ social media accounts to justify deportations, which is the sort of thing that might also give a much wider group of people reason to keep quiet online. More people have more reasons to withdraw to smaller places, or to abstain altogether.

But also, everything’s just TikTok, now

When TikTok showed up in the United States, all the other tech companies went sort of nuts. Here was a fast-growing platform, the first in many years, built from largely familiar features: A vertical feed; followers; visible metrics; short video. What made it compelling, though, was how much it relied on algorithmic recommendations. You could see videos from people you followed, of course, but most of what most people saw was stuff that TikTok was showing them based on past behavior, and the platform’s best guesses at what they might want to watch next. TikTok was still full of people, but most of what you saw was from people you didn’t know. Likewise, it was an unusually easy place to grow an audience, but one composed of strangers rather than friends. In that sense, it was hardly a social network at all, at least in the “connect the world” sense that the term had been used in the decade prior. It was more explicitly a platform for allocating attention and doing commerce, a user-generated entertainment product with a store attached.

Anyway, everyone copied it as fast and thoroughly as they could, in part because it was just a more fully realized version of what platforms like Instagram were already becoming, but mainly because it was a threat: new, foreign, and rapidly catching up. As an attempt by Meta, Google, X, and Snap to get TikTok under control, it didn’t work — that took multiple interventions from the federal government and some friendly billionaires — but in terms of engagement, TikTok’s competitors did pretty well by copying it. Previously social feeds filled with video made by strangers, resulting directly in the scenario described above by Mark Zuckerberg: Formerly social networks now filled with, basically, little shows and marketing, content from influencers, and random recommendation slop, an increasingly amount generated by AI.

They’re getting automated (and replaced?) by AI

As with, say, the labor market, it’s hard to say exactly what sort of effect AI is having on social media. But as with, well, much of labor market again, it’s pretty easy to just look around on social media and notice: AI is very, very visible. Facebook is full of AI-generated content, but it’s not uncommon to encounter AI-filled accounts on other social platforms as well. Social media executives are fantasizing about a future where generated content merges seamlessly with other things people might want to post, supplementing online friendships and providing entertainment against which to sell lots of lots of advertising.

For now, though, AI’s relationship to social media is a disaster only partly redeemed by irony: Companies like Meta, which are pledging to spend hundreds of billions of dollars developing AI tools, are seeing their nominally “social” platforms inundated by inhuman slop. Meta recently launched Vibes, a feed for AI-generated content, while OpenAI welded a TikTok clone to its image generator. And hey, maybe they’re a glimpse of a reconstituted social media future! Until then, popular AI tools seem almost perfectly engineered to undermine the social media platforms, glutting them with commodity content, breaking down their last remnants of community and sociality, and at the same time drawing users’ attention elsewhere, into chat interfaces that are doing their best to provide the sorts of pleasure and distraction that social media has been monopolizing for years, and making progress in that direction. Why risk asking other people for advice when you could ask your little friend in the computer?

Their valuable “network effects” are at risk of falling apart

The bull case for the social platforms is that, while they may have become entertainment platforms, they’re still dominant, and other forms of entertainment are in relative decline. They might not have a lot of room to grow in terms of usership, but they can compete in the more conventional manner of entertainment companies, combining programming with structural advantages and monopolistic tendencies to ensure, if not massive growth, rising profitability.

But the ability of TikTokified, hollowed out, slop-filled social platforms to continue to maximize engagement for now conceals a risk: that they lose the thing that helped make them some of the greatest businesses of all time, and that has protected them from decline in the past. Platforms like Facebook are clear examples of the power of network effects, through which the product is made better the more people that use it. A “social” network is better if everyone is there, both for users and the people who own it. These effects don’t just help draw people in — they help keep people once they’re there, even if utility and enjoyment has started to wane. Again: everyone’s still there.

It would be narrowly correct to say that “everyone” is still on social media, but not necessarily in the senses that matter most for keep people locked in. If users are lurking rather than posting, then they become invisible to one another; if content is commercially produced rather than personally and contextually created by users, it’s less tied to a specific network; if bad AI content is too visible, users might get the impression that the networks are becoming zombified; if they feel surveilled and constrained from speaking, they hardly feel like part of a network at all, much less a social one. A social media platform without network effects is just another source of endless video – another channel to flip to, or away from.

And for some of social media’s core uses, people have already moved on

Before they were worried about TikTok, every social media platform was worried about losing users to messaging, a straightforwardly social and far less monetizable means of communication that was becoming massively popular in the early 2010s. They resolved this worry by either building messaging into their platforms or by, in Facebook’s case, spending 19 billion to simply buy WhatsApp.

In the same Pew survey that tracked a recent decline in young peoples’ use of virtually every social media platform, WhatsApp was the one exception, seeing a late and dramatic increase despite being nearly as old as Facebook itself. The continued rise of chat apps, whether it’s platforms like Discord or more conventional messaging apps like WhatsApp, Telegram, or Signal suggests as strongly as anything that at least some — in fact, billions and growing — may be considering spending time in places where they have some semblance of control, privacy, and recognizable interpersonal dynamics. The group chat was with us before social media, and I’d bet it’ll be here long after it’s gone.

---

From Intelligencer - Daily News, Politics, Business, and Tech via this RSS feed