On what platforms became, what they're losing, and what neither AI nor social media can actually replace.

Social media used to be a place where people went to socialize and be entertained. Funny posts, political satire, photos from someone's weekend. It wasn't serious. It wasn't supposed to be.
Then it became a place where people made money.
And that changed everything. Not immediately. Gradually. But once the economics shifted toward engagement, the content shifted with it. Because engagement isn't neutral. It doesn't reward the interesting or the true or the thoughtful. It rewards the reaction. And some reactions are much easier to trigger than others.
If you've spent any time on X recently, you've probably noticed the patterns. Deliberately wrong information posted just prominently enough that someone will feel compelled to correct it. The correction generates replies. The replies generate more engagement. The original poster wins regardless of whether they were right.
Or the provocation format. "I think people who do X have something wrong with them." It doesn't matter what X is. The point is the framing, which is designed to make a specific group of people angry enough to respond. And they do. Because the algorithm doesn't distinguish between agreement and outrage. Both count.
Or the opinions that sound bold but are really just tribal. Not "here's a nuanced take on a complicated issue" but "here's which side I'm on, stated as aggressively as possible." These posts don't generate discussion. They generate sides. And sides generate sustained engagement over time, which is exactly what the platform needs.
This created a cycle that fed itself. Platforms rewarded the most extreme and reactive content. Creators learned to produce it. Audiences were served more of it. And gradually, the space that used to feel like a town square started to feel like something else. Angrier. More exhausting. Less like somewhere you went to connect and more like somewhere you went to argue with strangers.
The algorithm doesn't distinguish between agreement and outrage. Both count. That's the whole problem.
Trust collapsed along with everything else. Written and visual media had already lost significant credibility over the years, so people had turned to platforms like X for news. But the same dynamics that made the feed toxic also made it unreliable. Misinformation spreads faster than corrections. Manipulation accounts are indistinguishable from real ones until they aren't. The place people went to find out what was happening became a place where it was increasingly unclear what was actually true.
The numbers on AI usage tell an interesting story. People aren't just using AI tools for work. They're having conversations with them. Long ones. About news events, about personal decisions, about things they used to Google or used to ask friends or used to scroll through Twitter to get a sense of.
I think I understand why. What AI offers is something social media stopped offering a while ago: a response that isn't trying to win. There's no outrage optimization. No engagement metric. You ask a question and you get an answer, not a performance. You can push back and the system doesn't get defensive or aggressive. For a lot of people, that experience of algorithm-free interaction is genuinely unfamiliar at this point. And it feels like relief.
The control is yours, which is something social media platforms have been systematically removing for years. You decide what you want to know about. You steer the conversation. Nobody's feeding you the next thing before you've finished processing the last one.
Does this mean social media is dying? Probably not soon. But I do think we're watching something shift. The useful functions of these platforms, connection, information, entertainment, are getting harder to access underneath the noise. And people notice, even if they don't always leave.
Here's the thing though. Social media at its worst is a hall of mirrors that makes you angrier and more confused. AI at its best is still an extremely sophisticated text interface. And what's missing from both of them is the same thing.
Other people. Real ones.
When you take advice from someone you know, you know their track record. You know how their previous decisions turned out. You know what kind of person they are, not just what they think about this particular topic. You can watch their face when they're uncertain. You can tell when they're genuinely confident versus performing confidence. And if they're wrong, there are real consequences to that relationship.
None of that exists in a feed. And none of it exists in a chatbot either, however sophisticated.
The thing about social media is that it changed us in ways we could at least partially see. The rage, the anxiety, the difficulty concentrating, these were visible side effects that people could name and debate. We wrote articles about doomscrolling. We made jokes about Twitter brain. There was at least a cultural conversation happening about what these platforms were doing to us.
AI is different. The interactions are calmer. The experience is often genuinely useful. There's no obvious reason to be worried. And maybe there isn't. But we don't actually know yet what spending years in conversation with a system that always has an answer, never gets tired, and doesn't push back in the way that real disagreement pushes back, does to a person's relationship with uncertainty, with patience, with other people.
Social media showed us some of what it was doing. We could feel it happening.
What AI is doing to us over time, that's still forming. We're living inside the experiment and we don't have the results yet.
Related: The Algorithmic Cage