Some word vomit on writing
It doesn't have to be pretty. It doesn't have to be good. It just has to be.
One thing nobody tells you about AI drafts: the first one is often fine. That’s the problem. “Fine” is not what good communications looks like.
Good communications has a point of view in it.
A voice.
Something that sounds like a person who has actually thought about this. Fine doesn’t do that.
As a junior PR, you’re working hard, judged often on the strength of your writing. You hone your skills over time.
With LLMs taking this task and running with it, a key issue is that you’re removing the writing process.
When you write, you aren’t just trying to put words on a page. You start out with something - an idea, a goal, a single sentence, and you start to shape it as you write.
As you craft sentences, paragraphs, narrative arcs and structures, you’re actually shaping how you think about the topic.
It’s quite beautiful, in fact, when you read a piece of writing and you can almost sense the personality of the piece. It doesn’t matter if it’s a personal essay, a poem, or even a corporate byline on IT security.
Good writers start to infuse their writing with personality.
LLMs rob us of that.
They may give us back time and be efficient and help us get to where we’re trying to go... but it’s fundamentally missing the point of writing in my view
AI is very good at telling you what other people have said about a topic, and even what it thinks you want to say on the topic.
But it’s not giving you your own voice. It’s not creating your style. It is, at its simplest, spitting out the next most probable word, sentence and idea.
It’s also not very good at telling you what you think about a topic.
AI can spit out 10,000 word articles, it can give you a detailed research summary. And it can generate Amazon’s six page meeting notes so that you can move quicker.
But that fundamentally misses the point.
If the LLM does the work, and you don’t critically engage with it. That is to say, read it, understand it, interpret it and make it your own, then you’re not doing anything other than creating content for the sake of it.
If you take its output and don’t engage with the content properly before you share it or publish it, you’re not really doing much more than copying and pasting someone else’s work.
The skill of writing is fucking awful and hard.
Staring at a blank page is genuinely soul destroying.
Using LLMs to get some ideas to get started is great. But try and work through this idea yourself. Just try and write for 10 minutes.
The end goal is not supposed to be a perfectly written prose or essay. That comes with time. It’s about the journey on the way there.
You learn over time what audiences are engaging with and sharing. You see the pieces that get more reads than others. You look back and you analyse why and you try and replicate it.
And that’s how you find your audience. Through something that comes from you. It doesn’t need to sound smart. It’s doesn’t need brilliant grammar.
Formatting doesn’t matter. If it feels and sounds human, it will happen.
It won’t happen overnight. It will happen over time. And that means you do need to be consistent. If you’re trying to get fit, for example, you don’t just go for a run once a month. We all know (I hope) that you need to exercise at least a few times a week to see the effects. Same with writing.
You don’t just get good one day. You write badly.
As I finish writing this, I’ll say that this entire thing was my brain pouring onto the page. There has been no thought to finding links to back up anything I say or any thought of going back and editing it.
It’s my stream of consciousness onto a page.
If you take anything from this, it is to say that people will want to read what you write.
And if you use Ai to help that’s ok. What’ not ok is for it to be written by AI.
“If you didn’t care enough to write it yourself, why should I care enough to read it?”
If you’ve got this far, then let me share my secret about this post…
I used Claude.
I was reading an interesting article on how to grow on Substack and it talked about how to use notes. So I asked Claude, hey this article that i recently wrote. Help me turn it into a series of notes for substack, that I’ll post 1-2 x a day.
And it spat out 12 ideas for notes. Three of them I thought were alright. So I copied one and pasted it into Obsidian as an idea for later.
I then copied and pasted the other two. And then, I started writing.
You see, the three post that it spat out were ideas that I thought were interesting. But they weren’t well written, and not quite what I thought. And so I was just going to rewrite them. And I ended up writing this 1,000 word piece.