Vanta Logo
SPONSOR
Automate SOC 2 & ISO 27001 compliance with Vanta. Get $1,000 off.
Up to date
Published
3 min read

Trevor I. Lasn

Staff Software Engineer, Engineering Manager

The Internet is Becoming an Ocean of LLM-Generated Junk

The internet’s full of content, but most of it is becoming junk. I’m talking about the stuff generated by Large Language Models (LLMs). These AI tools are cranking out endless articles, and the quality? It's bad—really bad.

The internet has changed. It’s become flooded with content. Most of it is just regurgitated word salad or plain junk. The rise of Large Language Models (LLMs) is partly to blame. But here’s the catch—most of it is worthless. And I’m not saying this lightly.

When you read LLM-generated content, something feels off. It’s usually repetitive, overly wordy, and lacks depth. Sure, it can be grammatically correct and sound “professional.” But does it add real value? Most times, no.

LLMs don’t understand the content they generate. They’re just parroting back patterns they’ve seen from massive datasets. It’s like having a parrot that learned how to mimic conversations. Yeah, it can say things that make sense, but it doesn’t understand what it’s saying.

I’m Skeptical of Almost Everything I Read Now

Here’s the frustrating part: I’m skeptical of almost everything I read online now. I wonder, “Was this written by a human or generated by an LLM?” It’s not just that the quality is poor—it’s that the trust is gone. Even if the content seems polished, I still second-guess its accuracy and depth.

I’m sure you’ve felt the same way. You’re reading an article or documentation, and it feels oddly familiar. It’s like you’ve seen the same phrasing in three other articles. You start questioning, “Is this just recycled LLM output?”

The problem is that LLM-generated content isn’t just clogging up blogs or listicles—it’s starting to leak into everything. From tutorials to documentation, I spend more time verifying whether what I’m reading is trustworthy. And that’s time I could be spending learning or building something, not playing detective.

It’s getting harder to tell whether something is written by a human or a machine. But there are some telltale signs. If an article feels like it’s dragging on, or if it keeps repeating itself without adding any real value, that’s a red flag. LLMs tend to produce a lot of fluff to make the content seem longer or more thorough.

Another sign is repetition. If the same points keep popping up in slightly different wording, you’re probably reading machine-generated content. It’s like the AI doesn’t know when it’s already made a point, so it just keeps going in circles.

Here’s my advice: If you’re writing content, don’t just scratch the surface. Provide depth, real-world examples, and explanations that go beyond the basics. Otherwise, you’re just adding to the growing ocean of junk.

Note: This article is me blowing off steam. I don’t have any solutions to fix the issue.

If you found this article helpful, you might enjoy my free newsletter. I share developer tips and insights to help you grow your skills and career.


More Articles You Might Enjoy

If you enjoyed this article, you might find these related pieces interesting as well. If you like what I have to say, please check out the sponsors who are supporting me. Much appreciated!

Tech
5 min read

Pkl: Apple's New Configuration Language That Could Replace JSON and YAML

A deep dive into Pkl, Apple's configuration language that aims to replace JSON and YAML

Nov 1, 2024
Read article
Tech
11 min read

Google's Journey: From Search Engine to Tech Giant

Exploring the key innovations and strategies that transformed Google into a global technology leader

Oct 1, 2024
Read article
Tech
3 min read

When Will We Have Our First AI CEO?

Welcome to the future of corporate leadership. It's efficient, profitable, and utterly inhuman

Nov 4, 2024
Read article
Tech
5 min read

Is Age Really a Factor in Tech?

Silicon Valley has a reputation for youth worship. The 'move fast and break things' mentality often translates to a preference for younger, supposedly more adaptable workers.

Oct 8, 2024
Read article
Tech
5 min read

Can OSSPledge Fix Open Source Sustainability?

The Open Source Pledge aims to address open source sustainability challenges by encouraging companies to pay $2,000 per developer per year

Nov 17, 2024
Read article
Tech
4 min read

Sentry's LLM Integration Makes Error Debugging Actually Smart

How Sentry.io is using Large Language Models to transform error debugging from mindless stack trace reading to intelligent problem-solving

Nov 24, 2024
Read article
Tech
4 min read

No, Quantum Computers Won't Break All Encryption

Symmetric encryption algorithms like Advanced Encryption Standard (AES) are largely quantum-resistant already

Oct 31, 2024
Read article
Tech
5 min read

Why I moved from Google Analytics to Simple Analytics

How I ditched Google Analytics bloat for a privacy-focused analytics tool that bypasses ad blockers

Nov 9, 2024
Read article
Tech
3 min read

Introducing the Legendary Programmer Hall of Fame

Meet the innovators who laid the foundation for modern computing. Their contributions span decades, creating the tools and concepts developers use every day.

Oct 29, 2024
Read article

Become a better engineer

Here are engineering resources I've personally vetted and use. They focus on skills you'll actually need to build and scale real projects - the kind of experience that gets you hired or promoted.

Many companies have a fixed annual stipend per engineer (e.g. $2,000) for use towards learning resources. If your company offers this stipend, you can forward them your invoices directly for reimbursement. By using my affiliate links, you support my work and get a discount at the same!


This article was originally published on https://www.trevorlasn.com/blog/the-internet-is-becoming-an-ocean-of-llm-generated-junk. It was written by a human and polished using grammar tools for clarity.