LLMs and Content Repositories

The Google Search Bar / Screen is the most visited website in the world. We go there every day looking for answers. A search query, and we get a list of websites, the first 10 on the first page of the results. We go to the first one but may end up not being satisfied, we change the query until we get close to what we want. The feeling is that of skimming through a large book and hunting for a paragraph that can help you find the answer to your search query.

But the world changed overnight.

Chatgpt.

Instead of trying to find a page or a paragraph in a large book with a search query, you were now asking a question and getting an answer. The answer may be a hallucination, but it still does look real.

Each of us had many books, especially the ones like Encyclopaedia Britannica, prior to Google taking on our lives. Content repositories moved online, and we were paying for access to such repositories. It was in the form of subscriptions to online text, online videos, and online audio. All of these are great when you had the time to read through a lot of content, listen to an entire audio podcast or watch an entire video — in your quest to learn something new or find something you are looking for.

But, what do you do when you want quick answers, but don’t have time to read, watch or listen?

This is the hole that Chatgpt or similar solutions aim to fill.

Did you hear about what happened to Chegg? More than anything else, it was known as a website to submit your homework questions and get answers to them. And it was known only for that. Will you buy a subscription to only this one?

There is Udemy, a collection of video content for learning anything imaginable. But each video series is at least 30 minutes or more, and you may not even be sure if the content contains the answer to your question. Will you buy a subscription or buy a single course?

Chatgpt or similar LLMs have been trained on a lot of content. You ask a question and get an answer, it may or not be right. But, you don’t have to search through many results or go through a video. And Chatgpt is not limited only to just one form of content. These solutions have a free version which gives you almost all that you want.

So, why would anyone want to pay for content? Especially, when the content is not sufficiently differentiated or unique. Content was already a commodity. What will LLMs do further to it?

LLMs will keep improving, getting trained on more content, ultimately leading at least some of these to fold.