AI-generated content is flooding the internet

Day 31 / 366

LLMs like GPT-4 are trained on text from all over the internet. This is the text that was written by us humans, for the most part. When LLMs came, the promise was that it would replace humans in the workplace, and that would also include work that goes into creating content for the internet.

And people took that idea and ran with it. They created blogs and videos and images in bulk easily and flooded the internet with it. This AI-generated was not better necessarily, it was just stupidly easy to produce.

There are estimates that around 10% of all content on the internet is already AI-generated. And experts say that by 2026, that number can go as high as 90%.

In some cases, AI content is indeed useful. For instance, Amazon started showing an AI-generated summary of all user reviews. But the worst part is many of those user reviews are AI-generated as well. It has become so easy to spread disinformation using AI.

Lots of people on Twitter use AI to automate their tweets. When you are new on Twitter, a good way to increase followers is to reply to tweets from other popular profiles. So people can easily write a script to read the person's tweet, get an AI-generated response, and post it. Ultimately even the original tweet would be AI-generated, and we would just be left with AI’s talking with each other.

My opinion is that this is not a good use of AI. We cannot just get rid of human input, at least not yet. The internet will get filled with AI-generated spam and it will become more and more difficult to find useful information that you are looking for. Moreover, LLMs like GPT-4 have biases, that corporates have put in them. So the content generated by them will have the same biases as well.