Ask HN: Is anyone still resisting the slop onslaught?
Even as recently as a month or two ago his site seemed, to me at least, to be one of the final bastions of the public internet without a significant portion of its content being AI generated. Now, it feels like at least half of the articles I click on are full of LLM-isms and otherwise obviously AI generated. In spite of this, the comments are full of enthusiastic agreement or ardent dissent. Out of dozens of comments, none bother to acknowledge the inauthenticity of the content.
Is it finally time to throw in the towel accept the inevitable? Is it enough that the topic of an article or blogpost is interesting, even if the text itself is a pile of tokens? My work requires a lot of original research, and I'm putting a lot of information online for the first time. I'll run my website until traffic goes to zero. The work is important and I care. The big change is that traffic from the website covered my bills. Eventually, I'll have to earn a living through other means, and that means I have less energy for it. I've been seeing lots of sites ban AI content, so it's not just you. At the very least, I think there should be a law that requires AI content, videos, etc. to be identified as AI. It's getting really hard to tell what is even real and what isn't now :( If you are unaware, HN introduced a new rule because of the slop onslaught (onslopt?) https://news.ycombinator.com/item?id=47340079 It will take time for the hype to wind down still. Still a lot of people thinking they can plant a money tree because code is "easy" now, but it was never really the hard part anyway. No. I keep going because I still care. First off, I don't get the "obvious AI generated" take. OpenAI models (which are the most used) have some patterns which they like to repeat, likely due to aggressive fine tuning, but this is not inherent to LLMs or AI. Claude, Grok, and such don't have this. Only "keyboard-inconvenient but correct punctuation" in the form of em-dashes. Most of the AI content I see is fine. It's people using AI as a tool to create something nice. Those blog posts are perfectly pleasant to read, and there's plenty of AI artists who create very good art. AI is a tool, the quality depends on the person who uses it. What is annoying is when people just hook up an LLM to an account, and have it post autonomously. An LLM is just a statistical model, it has nothing to contribute to a human discussion. That's actual "slop"; people generating and posting without effort. The anti-AI luddites are not nice either. They'll accuse some perfectly fine work of being AI, whether it is or isn't, and become hostile. They don't participate in good faith, and they often repeat false information (or just lie) to suit their narrative. These people are as bad as, if not worse than, the people letting their ClawdBot loose on a comment section. I flag all "built $X with AI" posts, and downvote all comments on them. For some reason "vibe coding" irritates me as does slop. Personally I have always spent more time than I would have liked online. AI slop degrades the quality enough that it makes it easier to cut back on that wasted time. I’ve noticed this same effect. I find myself reading more books now that the internet has become a place I no longer recognize or enjoy.