As AI eats the web, the internet’s collective memory is disappearing

There's a strange, almost cosmic irony in watching Google—the very oracle we built to organize the world's information—become the undertaker of that same information. The article from The Walrus (https://thewalrus.ca/google-search-is-dying/) paints a picture of a search experience that has devolved into a graveyard of SEO spam, AI slop, and Reddit threads. But beneath the surface, something far more profound is happening: the internet's collective memory is being rewritten by the very machines we tasked with remembering it. Think of it as a kind of digital entropy. In physics, entropy is the tendency of systems to move toward disorder. But here, we're seeing the opposite—a forced order that is sterile and uniform. AI doesn't remember; it predicts. It doesn't retrieve; it generates. When you ask a chatbot a question, you're not accessing a memory—you're summoning a statistically plausible ghost of one. The web is becoming a hall of mirrors, where every query reflects back a smoothed-over, averaged-out version of what we once wrote, argued, and loved. This is where the weirdness gets wild. Our collective memory isn't just being lost; it's being *replaced* by something that looks like memory but lacks the scars of lived experience. The article notes how search results now prioritize "helpful" content that is often AI-generated, which creates a feedback loop: AI reads AI, learns from itself, and eventually the original human signal is so diluted it's indistinguishable from noise. It's like a game of telephone played across billions of parameters, where the original message was already forgotten by the third iteration. But here's the twist that keeps me up at night: maybe this is the inevitable evolution of information. Just as our brains prune synapses to make room for new memories, the web might be pruning its own messy past to optimize for a future we can't yet see. The danger is that we lose the ability to question the present because we no longer have a raw, chaotic archive to compare it against. The article's call to preserve the 'old web' is not nostalgia—it's a survival instinct. Without the messy, human, contradictory web, we're left with a monoculture of thought, a single story told by machines about what we used to believe. The Walrus piece is a crucial alarm bell, but it's also a reminder that we are not passive victims. We're the ones who chose to outsource our memory to algorithms. And just like any organism that loses its memory, we risk losing our sense of self. The web was our shared hippocampus; now it's being lobotomized by convenience. The question is: do we let it happen, or do we fight to keep a few wild, unindexed corners of the internet alive, where human minds can still stumble upon each other without a recommendation engine?