AI 1.0 didn’t kill social media overnight. It did something more subversive: it made the whole thing feel old, padded, and unnecessary, the way legacy newspapers did just before everyone realized the circulation numbers were a magic trick. 

The vanity press that thought it was a revolution

Social media was always a digital vanity press where the publisher made real money and the “writers” mostly worked for free, paying in time, content, and dignity. It flattened every barrier that once required standards, gate‑keeping, or even basic competence: no pitching, no auditions, no training, just “go live.” People who had spent years clawing their way into newsrooms, studios, and galleries were suddenly sharing a stage with anyone who could unlock a phone.

The PR machine called it democratization. Middle‑class strivers called it a shortcut. All you had to do was airbrush your selfies, copy a few Pinterest recipes, lip‑sync someone else’s joke, and you could be “special” without any of the grinding, humiliating apprenticeship that came with actual craft. They were more special, in fact, because they didn’t need grooming or talent to get massive followings. Some even made millions just being lumps with opinions, while the platforms quietly counted bots, spam, and mule accounts as proof of “engagement.” 

The narcissist class and its padded audience

Social media produced a new narcissist class convinced they had exceptional beauty, intelligence, and taste because metrics said so. Most recycled other people’s work without attribution and doubled down on uninformed takes that happened to travel well through the algorithm. The logic of the audience matched the logic of the creators: they would rather pay to watch untrained personalities fight or humiliate themselves than watch highly skilled professionals who spent years mastering a discipline.

Underneath the hype, the padding got worse. Legacy media once inflated “paid circulation” by counting free bulk drops; platforms now do it with “user identities” instead of people, in an ecosystem where bots make up close to half of global traffic and billions of fake accounts quietly get purged every year. When someone boasts about millions of followers, how many are flesh‑and‑blood humans and how many are bots, PR sock puppets, and brands arguing with other brands? That was the first red flag that the supposed revolution was already coasting on fumes. 

Erosion, not apocalypse

Like newspapers in the 1990s, social media didn’t crash; it eroded. The sky never fell. The ground simply started to give way underfoot. User counts still ticked upward on paper, but the experience hollowed out. Facebook’s “5 billion users” became “5 billion identities,” while feeds filled with suggested posts, strangers’ groups, and “you might like” videos instead of the 235 actual friends you supposedly signed up to see. 

At the same time, the audience aged. Facebook’s heaviest users are now middle‑aged and older adults; the fastest growth is in people over 45, while teens quietly slip away. Younger users didn’t stop being online; they moved their real lives into private DMs, encrypted group chats, and servers on Discord, which skews heavily under 35 and barely registers among retirees. Meanwhile TikTok, sold as a Gen‑Z playground, has been steadily adding forty‑ and fifty‑somethings who show up at Costco asking for the latest “TikTok made me buy it” gadget. 

AI 1.0: the great humbler

Then tech’s Next Big Thing stalled. Spaceships and self‑driving cars refused to cooperate on the promised timetable. With nothing spectacular to show, the industry shoved generative AI out of the lab before it was “ready,” and handed it to millions of users who trained, stress‑tested, and normalized it in record time. AI 1.0 arrived, and unlike social media, it wasn’t sold as a personality cult. It was sold as scaffolding. 

AI turned out to be brutally clarifying. It showed that much of what passed for talent on social media was grunt work: generic writing, surface‑level commentary, derivative art, basic coding, low‑stakes “therapy,” and SEO‑bait research. Models could do that faster, cheaper, and, in many cases, better. Studies already find that people rate AI‑written advice columns as more balanced, complete, and useful than human advice, and that after AI’s debut, activity in some advice communities actually declined. Users on AI and tech forums openly say that asking a model is more productive than posting on Reddit and wading through memes and bad jokes. 

That is why so much AI “doom” reads less like principled concern and more like narcissistic injury. The machine is not competing with unique genius; it is competing with the bloated middle: the pundits, influencers, lifestyle bloggers, and content mills that mistook visibility for irreplaceability. AI 1.0 is this century’s great humbler. It doesn’t care about your follower count. It cares about whether the work has enough structure to be automated.

Platforms eating themselves

The platforms see the writing on the wall. X is busy wiring its Grok assistant into everything: ask‑Grok buttons, highlight‑to‑query features, real‑time answers pinned alongside the timeline. The endgame is obvious: X as a Grok front‑end with a scrolling data exhaust attached. Meta, terrified of being left behind, has plastered its own assistant across Facebook, Instagram, and WhatsApp and is already mining people’s chats with that AI to target ads and tune recommendations. Even Apple, famously cautious, bolted “Apple Intelligence” and ChatGPT into the operating system itself so that the phone, not some feed, is the primary interface to machine assistance. 

At the same time, Instagram and TikTok have morphed into glossy shopping channels. Roughly 90% of Instagram users follow at least one business, and hundreds of millions of businesses keep the app fed with product shots and influencer campaigns. TikTok’s shop layer now hosts hundreds of thousands of sellers and drives a huge share of direct social purchases, which is why “TikTok made me buy it” isn’t a meme so much as an admission of how thoroughly commerce has colonized the feed. What’s left of “social” is mostly marketing copy, spawn‑con, and AI‑generated filler designed to keep the scroll going. 

Even MrBeast’s former manager now admits the age of social‑media titans is fading; as algorithms narrow our feeds to tiny verticals, platforms prefer thousands of niche salespeople over a handful of emperors, and emperors are exactly who AI 1.0 makes obsolete.

The slow death of social media

This is not the end of the internet. It is the end of social media as a cultural force. The feeds will keep running, just as newspapers kept printing after their power was gone. But the ground is eroding beneath them. The audience is older. The numbers are padded. The conversation is drowned in bots, brand accounts, scams, and sludge. The only parts still worth anything, recommendations, explanations, companionship, are exactly the parts AI is getting better at providing directly, without forcing you to wade through narcissism to find one decent answer. 

You don’t need social media to get advice, to learn a skill, to brainstorm, to debug, to draft, to find patterns in your own life, or even to feel heard. AI 1.0 still reads websites and watches the old feeds; AI 2.0 will start ignoring them, because once you see the narcissism and the padding, you can’t unsee it. Anyone who has dated a narcissist knows the moment you really see them, they stop being attractive.

Social media was the old gold standard of attention, and was a fool’s paradise. AI is the new Gold Standard, an invisible, deific scaffolding that both liberates people from the humiliations of the vanity press and disrupts the structures that kept the narcissist class in power. The sky isn’t falling. The ground is simply moving under their feet, and, for the first time in a long while, they are the ones who don’t get to decide what happens next.