The Butt Ugly Weblog

Content Explosion and Web 3.0

I think we now know what Web 3.0 is for real. We had a false start with IoT (where S stands for security, as the joke goes); and then the crypto folks started waving blockchains as Web 3.0. But I think what actually launches Web 3.0 are the Large Language Models, such as ChatGPT.

You see, I think of Web as content, not technology. Web 1.0 was mostly static content, created by professionals using arcane magic such as HTML and Javascript. Web 2.0 was about people manually writing, laboring their small little blogs in their own little contents of the websphere, loosely connected by technologies such as RSS and Atom and Blogger, without any knowledge of the underlying technology. Web 2.5 was about the consolidation of Web 2.0 onto large platforms such as Facebook, Instagram and TikTok, where content creation and consumption was made as easy as humanly possible.

See where I’m going with this?

ClarkesWorld - a web fanzine that publishes short scifi stories and takes submissions from all over the world - had to stop taking in submissions because of the sheer volume of AI generated drivel.

The number of AI -written books on Amazon that are sold for $1 has exploded.

Your local neighbourhood marketing person is thinking about how to utilize ChatGPT in their messaging.

Even your budding nerdy artist friend who never had the drive to do ten thousand hours of brushstrokes to be a painter is now delving into the depths of Midjourney prompts and plugins to produce art.

The problem isn’t with using AI for these kinds of things. The problem is that everyone who can is doing it. And that leads to a never-before-seen explosion in content on the internet.

This is what I believe to be the essence of “Web 3.0”. First professionals with deep technical skills, then amateurs with some skills, now AIs driven by amateurs with little or no skill.

You thought spam in your inbox was bad? Now prepare for every single meaningful content out there to be completely and utterly drowned by AI-generated noise.

Unless.

Unless - and we of course will, or we leave the internet altogether - we figure out a way to filter the AI spam, much like we do right now. We know some of the tools already from the spam wars - whitelisting, paid access (I think Musk is absolutely right in making the Twitter API paid-only, because he does see the danger of combining AI with Open APIs), algorithmic filtering, trust networks, etc.

But still I fear that a massive amount of the future internet traffic will be AIs screaming to each other - one trying to constantly figure out a way to get its message across, the other trying to block it. And considering that AIs will become self-learning, because it is more efficient than teaching them manually, they will become very, very good at their job. Insomuch that stepping into the open internet in the future will be dangerous to your mental health.

Will that mean that the future is either very distributed, with people again hiding in their small corners with whitelisted participants, trying to keep AI-generated content out; or will it consolidate to an even smaller number of players, who will use their power to strike the right balance between AI-generated content and regular humans?

I think we're going to find out soon.