LLM Spam: exploring LLMs' potential to litter the internet

The LLM Spam was a short-lived study on how LLMs will impact the internet as we know it. The experiment involved generating a Python script that would call OpenAI’s ChatGPT API to create articles. The goal was to determine if it would be possible to do this at scale, use clickbait tactics, and manipulate users into engaging with the articles. The answer is yes.

This was an extremely interesting project that didn’t teach me much about coding, but it was very thought-provoking regarding what the internet might look like in the future and the dangers of LLMs.

Some interesting facts:

• The whole project cost less than $25 in API credits
• 1,082 posts were automatically generated and published over the course of a few days
• 1.32M words were written
• The website received 2,712 visitors and 3,334 views
• Posts that provoked anger in visitors, generally when they contained misinformation due to ChatGPT’s hallucinations, were the best-performing posts

The code for this project is not open source as I believe that publishing it would be a net negative for the health of the internet.

Technologies: Python, OpenAI API, WordPress API