Generative AI And The Future Of Content On The Internet

If you have been on any social media over the past few years, you have probably heard people talking about “the algorithm.” For those who don’t know, “the algorithm” is essentially just what social media sites use to recommend content to their users. While every different social media has a different algorithm, they all have essentially the same purpose, to keep the user on as long as possible. Right now, they do this by gathering as much data as possible from the user, (think watch time, likes, following, etc), and then running it through a machine learning algorithm to determine what the user likes and what to recommend. This works very well, with many users getting addicted to the sites spending numerous hours a day on them. However, no matter how well the algorithm might work now, I believe that there is still a major bottle neck in the system. One that, if fixed, could potentially make social media way more addictive… The problem: The content has to be made by a human.

But wait… What’s wrong with humans?

There are two answers to this question depending on your morals:

• 1. Absolutely nothing. Humans have the special ability to be able to pour their heart and soul into a piece of content to tell a story or to share their emotions. In the same way that events were passed down through stories and literature before the internet, we are now able to do the same through films and home video. Having this be substituted by some soulless machine will take away the very meaning and point of a film. It would take away the very essence of what creativity is and replace it with a system only designed to gain your attention and profit off of your addiction.
• 2. Noooooo! It’s reducing watch time!!! Its not efficient enough :0 Must improooove

The point of a social media is to try to get you to spend as much time on the app as possible. They do this so they can show you advertisements and make money. To do this, they try to make an environment where you feel like you are getting some sort of value from it (entertainment, educational, etc). Right now, they do this by using their algorithm to select content it thinks you will get the most value from and shows it to you. This is great if the perfect content for the user already exists out there, but if it doesn’t, the algorithm will have nothing perfect to show to the user, and the user will get less value, causing them to spend less time on the app. This led me thinking:  “How could a social media give the absolute best, most optimised content to its users?”

And then it hit me… Generative AI!

Generative AI and hyper-personalised content

What if in the future, social media services don’t have to rely on humans to create content for you. Imagine if when you scroll, the algorithm will not only take note of the watch time and other metrics, but the BPM of the song, how clean the room in the video is, how the person is behaving, what hairstyle do they have, and so on and so on. It could then feed it into a Generative AI and it will create the perfect video for you. Not only this, but it could create highly personalised ads by having there be product placement in the background or by having the generated character wear or use the product. The way I see it working goes something along the line of this:

The app would have a similar interface to TikTok, with the main interface just being an endless scrolling feed. When the user first starts to scroll, they will only be shown user generated content, so the algorithm is able to get a sense on what the user likes. Then, as it learns, it will slowly start to inject the AI generated content seamlessly into the feed without the user knowing. If positively receipted, the AI generated content will start to take up more and more of the user’s recommendations until it consumes a significant portion, if not all of it.

While this may seem far off, we are already starting to see early glimpses into this technology. A few days ago, I was reading through Hacker News and I stumbled across a very interesting program. It’s a python script that uses GPT4 to create Wikipedia like websites on fake topics. It then creates links to other things that don’t exist. Then, when those links are clicked, it generates the content for that page. From there the user can read more and click on more links to the fictional content, and so on and so forth. What this does is create an entire fictional, AI generated universe for the user to explore. Even though it is just randomly generating content, it was still very fun to mess around with.

Why a system like this would be bad

With the advancement of generative AI like GPT4 and Text to image AI like stable diffusion, we are incedingly able to imagine up an endless amount of content on whatever we desire. However, no matter how harmless it may seem now, there would be numerous consequences if a social media like this was to be used in the future. Going back to the question of what is wrong with human content, I strongly agree with the first point. I believe that we already have gone past the point of the content we consume being good for us. Over the years I have been on YouTube, I have noticed that the content is getting shorter and less informational over the years. While in the beginning the content was made by passionate people just wanting to talk about what they love, now it has devolved into people doing anything to try to get your attention.

If we were to have a social media where it creates content to be exactly what the user wants, it would not only further add to the problem of the content we consume desperately trying to get our attention, but it would also create environments like echo chambers that would be more dangerous than what we have today.

Conclusion

As much as I believe that this technology would be detrimental to us, I still think that it will be inevitable that something like this will be implemented in the future. While I don’t think this blog post will do little to anything to stop it, I still wanted to make it to at least throw my hat in the ring and predict what’s coming. So, if you are a person reading this many years in the future and a technology like this exists, try reconsider your usage of that app.