AI Writing Isn't Content Boom, It's Traffic Self-Satisfaction

Just saw an article on Hacker News with a pretty blunt title: we can't let AI writing take over the internet.

The author made a jarring point: traffic from AI and bots has grown from a small fraction to over 50% recently, and he even worries it's heading toward 99%.

I work in enterprise AI, and these past few days I've been testing Coze templates and Feishu Bitable. The reality feels more grounded. It's fast for models to write something that "looks professional," but when you actually put it into a workflow, it doesn't understand boundaries and can't bear responsibility. Worse, humans can still observe the real world when writing, while machines just keep feeding themselves.

There are contradictions in the material too. Some reports say AI writing hasn't truly overwhelmed human content yet, but comment sections on platforms like LinkedIn are already full of templated long posts. Whether it has overwhelmed us or not might not matter. What matters is that people have started using synthetic content to validate each other, like an out-of-control jar of syrup—first it's a little sticky, then the whole street becomes impassable.

Can't call it a content boom; training data is starting to eat its own vomit. Before hyping "everyone is an author," let's get source labeling, accountability entities, and retrieval filtering done first. Otherwise, this so-called intelligent internet will just become a garbage dump that's better at self-validation. That's it.

---

📌 This article is compiled from Hacker News. Original text: https://www.derekthompson.org/p/the-internet-is-drowning-in-ai-slop

Copyright belongs to the original authors. This is a compilation and independent analysis based on public reports.