How do you write, and more importantly, how do you write blog posts?
For starters, one may want to try to keep to the subject matter, I think. In this post, I deliberately picked ‘How to blog post?’ because I wanted to write about it. I haven’t made that many posts on here yet, so it kinda makes sense that the first few posts are a bit about learning and setting out a foundation for later posts.
AI as a springboard
To learn blogging properly, I believe AI can act as a good springboard; it can give tips about how to structure a post or give links to tutorials related to whatever you are going to write about. As long as you don’t plagiarize the AI, it can be a great tool for learning most things, even writing blog posts.
I do not mean that one should avoid auto-generating text, but it should be clear in any text who the author is. This can also be a gray area, as sometimes you need to read up on a topic to understand it; whilst doing so, it can be easy to accidentally memorize a text without fully understanding it and thus end up partially copy-pasting what someone else has written. It is best if one is able to use their own words to explain something or, even better, artistically express it creatively in new ways that haven’t really been done before.
In iterations
One way to use AI as a springboard is to do so in iterations. In the first iteration (let’s call it iteration zero), one just asks the AI how a skeleton of a post is structured in very general terms. Now, there probably is no one-and-only kind of skeleton for blog posts, so the AI comes back with some new set of questions instead of direct answers. Then the next iteration is answering those questions yourself, which leads to some kind of new prompt where the AI eventually asks you to select a topic.
This is basically where the real work begins because now you have to choose something to actually write about. What I’m getting at here is that in each step of the writing process, the AI can iteratively be used to guide you in the direction the work should proceed in.
Refining the text
Once the work is finished, there is also another iterative process of refining the text; it seems especially important to “remove unnecessary text”. At least, that is what I’ve heard famous writers claim to be important.
If you do not remove unnecessary text, then you most likely have some text that makes the text more voluminous, and if it is too big with too little actual content, then humans will not be willing to read it, or it is going to be rather difficult to do so.
On the other hand, AI can read way more than humans, so maybe it is not as important to remove or refine text as it once used to be. If a human doesn’t find your text very easy to read, then that person can just ask an AI to rewrite it in whatever way they find good.
Limits to what an AI can produce
That said, there are limits to what an AI can produce. One would think that an AI can write anything, just like a human can write text in any fashion, but in reality there is a strict boundary for what AIs can or cannot do, even if that boundary is not entirely obvious.
LLMs use deep machine learning, and the thing about that way to learn is that you optimize for one single outcome, not a nuance of different outcomes. Meanwhile, there are more humans than there are LLMs, and each human can have their own optimized way of writing or multiple different ways.
Anyways, I will try to wrap this text up. In short, AIs are a great tool for helping to write text, but they shouldn’t write text for us. If we haven’t written something ourselves, then we may not even understand what is being written. Hopefully, this will help motivate both me and the reader to write more and not let AI do everything.