When AI fills a feed, likes keep rolling in but replies dry up. Is the platform still worth using?

This explores whether AI-generated content speeds up the cycle where platforms decay for their users while still making money, or whether it forces platforms to push back. The collection never uses the word 'enshittification', so there is no direct answer here. What it does have is a detailed account of how AI content wears down social platforms. The surprise is in where the damage happens. The standard story is about platforms squeezing users and advertisers for profit. The corpus describes a quieter loss: the social signals that made a platform worth using in the first place stop meaning anything, and revenue keeps coming in while it happens.

The central mechanism is false social proof. AI posts collect likes because they are comprehensive and confident. But they invite almost no replies or pushback, so you get visibility without any conversation behind it Why do AI posts get likes without inviting conversation?. Over time this pushes out human influencers. No one builds a lasting reputation, yet engagement metrics and monetization stay healthy Does AI content displace human influencers on social media?. That gap between 'the dashboards look fine' and 'the platform has stopped doing its job' is the condition decay cycles feed on. A related note argues the loss is conversational style itself: posts no longer address anyone or expect an answer. That happens below the level that content moderation, fact-checking or recommendation tweaks can reach Does AI threaten social media's conversational function?. The usual fixes can't see the problem.

The trend isn't uniform. Detector-based measurements show the share of AI-written posts on Medium and Quora rising from about 2% to about 38% between 2022 and 2024, while Reddit only went from about 1.3% to 2.5% Is AI-generated content rising faster on some platforms?. Platforms built around threaded conversation and community norms seem to resist much better than platforms built around standalone answers or essays. There is also evidence that platforms can push back on purpose. In one study of nearly 179,000 matched pairs of posts, the platform's algorithm gave AI-generated content less exposure than human content Can algorithmic distribution prevent AI content from overwhelming creator diversity?. That note covers only part of the study, so treat it as a lead, not a settled finding. Still, it suggests recommendation systems can act as a brake on the decay, not just an accelerant.

The broader economic framing explains why the brake is hard to keep pressing. If AI produces content faster than people can judge it, confidence in everything drops, much as money loses value in hyperinflation Can AI generate knowledge faster than humans can evaluate it?. The system keeps working only because readers stop checking. The corpus calls this 'cognitive surrender' and cites studies where about 80% of AI output was accepted without challenge When do users stop checking whether AI output is actually backed?. For platform decay, that means users may not notice the decline the way they notice ads or paywalls. The slide toward worse content can feel normal.

So the best answer the collection supports is this. AI content probably speeds up the decay cycle, but by a different route than the familiar one: it hollows out reputation and conversation rather than visibly extracting value from users. Whether a platform declines depends on two things. One is how conversational its design is. The other is whether its ranking algorithm chooses to favor human voices. For the wider picture of slow, incremental replacement wearing away human influence over systems that seem to run fine, see Does incremental AI replacement erode human influence over society?.