When AI-written posts collect likes with no real author or conversation behind them, do the numbers still show what people value?

This explores whether artificially amplified content weakens the signals people use to judge what others value, like engagement counts and visibility. The corpus covers this mostly through AI-generated posts rather than paid boosting or bot networks, so read it as being about engagement that no real author earned. The short answer is yes. The useful insight is how it happens: amplification doesn't just inflate the numbers. It separates the numbers from the conversation that used to give them meaning.

The key idea is that social proof was never only about counts. Historically a popular post also drew replies, pushback and back-and-forth, and that exchange is what made its popularity mean something. AI-written posts tend to be comprehensive and confident, so they collect likes while leaving little to argue with. The result is recognition without dialogue Why do AI posts get likes without inviting conversation?. A related note argues the bigger loss is conversational style, meaning posts that are genuinely addressed to someone, rather than accuracy or sentiment. That loss happens below the level that moderation or fact-checking can reach Does AI threaten social media's conversational function?. Over time this pushes human creators aside. Engagement keeps flowing and keeps being monetized, but it no longer builds any particular person's reputation, so platforms slowly stop doing their job of surfacing credible human voices Does AI content displace human influencers on social media?.

Why don't audiences notice and discount the inflated signal? People can't reliably tell AI content from human content. Across 30 studies, detection accuracy sits around chance Can people reliably spot content made by AI?. Even when people are told AI was involved, they become more skeptical, yet a third to more than half stay persuaded Does telling people an AI wrote something actually stop them from believing it?. So labeling content helps, but it doesn't bring back the meaning the signal has lost.

The less obvious finding is that this pattern isn't unique to social media. It shows up wherever people use a cheap surface cue to judge credibility. In AI search, answers with more citations win user preference almost as much when the citations are irrelevant as when they're relevant Do users trust citations more when there are simply more of them?. People also trust ChatGPT because it feels conversational, not because it's accurate Does conversational style actually make AI more trustworthy?. In academia, AI has produced hundreds of complete papers with made-up theoretical justifications, copying the outward signs of scholarly credibility at scale Can AI generate hundreds of fake academic papers automatically?. And a separate attack type hides promotional content inside model outputs, which is amplification you can't see at all Can language models be hijacked to embed hidden advertisements?. In each case, a signal that used to stand in for real human judgment can be produced without that judgment.

The corpus also points to a way out. In one study, readers given no information about where claims came from couldn't tell true statements from fabricated ones. When an interface showed them how many claims had been verified, they could tell the difference again Can readers tell truth from fabrication without evidence signals?. That suggests the fix isn't to stop counting likes. It's to show signals that are harder to fake, like whether real people actually replied or where a claim came from. One gap: the collection has little direct research on paid promotion or bot-driven boosting, so the evidence here comes mainly from AI-generated content.