The pinnacle of enshittification, or Large Language Models

The argument

A recent post on the Gentoo project's blog argues that much of the current debate around large language models (LLMs) has slid from technical critique into ritualized panic and marketing spin. The author says they are tired of reading about LLMs — not because the criticism is wrong, but because it’s often right. It has been reported that the post rails at two connected problems: the casual anthropomorphizing of models, and the way the industry — and broader culture — happily labels them “intelligence” while they mostly imitate us.

How the post explains LLMs

The piece walks readers back to basics. LLMs, the post argues, predict what token comes next given context; they don’t “comprehend” in the way humans do. Call it a parrot with a keyboard. It has been reported that the author insists these systems remix their training data rather than generate genuine understanding or novelty, and that their fluency makes that mimicry dangerously convincing. Some people, the post warns, now believe chatbots are conscious — a claim the author treats with alarm.

Why it matters

What’s the emotional core here? Frustration. The writer is furious that society’s appetite for easy narratives — hype, marketing, quick answers — has effectively trained models on what he bluntly calls “bullshit,” and then optimized them to produce more of it. The consequence, he suggests, is twofold: we get dazzling-sounding outputs that can be hollow, and we risk normalizing sloppy thinking. Sound familiar? It’s the same trend that has been nagging at tech culture for years, now amplified by machine learning.

The takeaway

The post is less a how-to and more a warning shot: treat LLM outputs skeptically, resist anthropomorphic shortcuts, and remember the models’ limits. It has been reported that the author urges readers to keep asking the awkward question — are we building tools that help thought, or tools that merely echo our worst tendencies back at us? Either way, the debate isn’t going away anytime soon.

Sources: blogs.gentoo.org, Lobsters