The human-only social network: feeds in the age of AI

Something is changing about feeds, and it is changing fast. The images are increasingly AI-generated. The text is increasingly AI-written. The video is increasingly synthetic. And at least one major platform has openly floated populating its feeds with AI-generated profiles, characters that post, comment and interact as if they were people. The feed, which began as a place to see what your friends were doing, is becoming a place where you can no longer be sure that anything, or anyone, is real.

This piece is about that shift and the new kind of value it creates: the assurance that what you are seeing is human. I run a network built entirely around real people, so I have a stake in this being valuable. I will make the case and, as ever, be honest about its limits.

What is actually happening to feeds

A few trends are converging, and they are worth naming plainly.

AI content is flooding in. Generative tools have made it trivially easy to produce convincing images, video and text at scale, and a great deal of it is ending up on social feeds, much of it low-quality, engagement-baiting material that has earned the nickname “slop”. The sheer volume changes the texture of a feed: more and more of what scrolls past was not made by a person at all.

Platforms are leaning in, not pushing back. Because AI content is cheap, endless and engaging, it suits the attention economy perfectly, so platforms have little incentive to keep it out and some incentive to add more. Notably, there have been moves to introduce AI-generated profiles and characters directly into feeds, which drew significant public backlash, an indication of how uneasy people are about it.

The “dead internet” feeling. There is a half-joking, half-serious idea, the “dead internet theory”, that much of the internet is already bots talking to bots and AI generating content for algorithms to rank for other AIs. It is overstated as a literal claim, but it captures a real and growing unease: that the human share of the internet is shrinking, and that you can no longer assume a person is on the other end.

Why “human” becomes valuable

When something becomes scarce, it becomes valuable, and human-made, human-verified content is becoming scarce. The assurance that a post came from a real person you actually know, rather than a model imitating a style or a bot chasing engagement, is a new kind of premium.

This matters most for friendship specifically. The whole point of seeing a friend’s post is that it is your friend, their real life, their actual face, their genuine update. An AI-generated image in a stranger’s feed is one thing; the prospect of not being sure whether your friend’s post is really your friend, or a synthetic interaction inserted by the platform, undermines the entire reason the feed existed. For a network about real relationships, “everyone here is a real human you know” is not a gimmick; it is the foundation.

The honest limits of the idea

A fair piece has to resist overclaiming, so here are the caveats.

“Human-only” is harder to guarantee than to assert. Verifying that every account is a real person, and that no AI content sneaks in, is a genuine technical and operational challenge, not a box you simply tick. Any network making this claim should be honest that it is an ongoing commitment and an arms race, not a solved problem.

AI is not the enemy in general. The point is not that AI-generated content is worthless or that AI tools are bad; they have enormous legitimate uses. The point is narrower: that a space for your real friendships should contain your real friends, and that mixing synthetic content and characters into that specific space undermines its specific purpose. Being pro-human-feed is not being anti-AI; it is being clear about what one particular kind of space is for.

And human content is not automatically better content. Real people post boring, unflattering, mundane things; AI can produce more polished material. The value of the human feed is not that it is higher quality but that it is true, the actual texture of actual lives, which is exactly the thing a friendship feed is supposed to show.

Where 142 fits

142 is built to be a human-only space, by design and by structure. Every connection is a real person you have actually added, capped at 142, with no public discovery, no creators, no brands, and no AI-generated profiles or characters; the feed is your real friends and nobody and nothing else. Because it is funded by subscription rather than engagement, there is no incentive to pad the feed with cheap synthetic content to keep you scrolling, which is the pressure pushing AI slop onto ad-funded platforms. We treat “everyone here is a real human you know” as a core commitment rather than a feature, while being honest that maintaining it against a rising tide of AI is an ongoing effort, not a one-time guarantee. The broader design philosophy is in why 142 has no algorithm and 142 vs Instagram. It costs £3.99 a month or £29.99 a year, and is launching UK-first; you can join the waitlist at 142.social.

Summary

As AI-generated images, video, text and even profiles flood social media, a new value is emerging: the assurance that what you are seeing is real, made by an actual person. Feeds are filling with synthetic “slop”, platforms have floated inserting AI characters into the mix, and a genuine unease is growing about how much of the internet is still human. A human-only network answers that directly, especially for friendship, where the entire point is that your friend’s post is really your friend. The idea has honest limits, it is hard to guarantee, AI is not the enemy in general, and human content is not automatically better, but for a space built around real relationships, being genuinely, verifiably human is the foundation, not a feature.