Writing, pouring my heart out and AI articles

Calendar icon

Published: 27 June, 2023

Clock icon

7 mins(s) read

Tags icon

Tags: #opinion

Among the things AI has become capable of, is writing well, detailed articles. Articles which can be written in different styles for different moods. As long as you prompt it appropriately. Under an year, you will start seeing visibly high output of articles which are published this way. There will be a high output of marketing and sales articles. Not to mention fake news and propaganda. It will make it harder for readers to differentiate between what is the truth and what is not. There will be people who use it to correct their grammar and typos. Then there will be people who are genuinely trying to make their writing unique with the help of the LLMs/AIs. The reason why I say this is, not all of them will be ill intended on the AI assisted writing spree.

As someone who writes blog posts on free time, I am not sure how to feel about it. This definitely adds accessibility to people who are not good at writing. It is also very helpful for when someone starts to write. Especially if they intent to write in English and it is not their mother tongue. Obviously we all know that using it when necessary is not bad and is in fact a good idea. But technology is often used in ways it is not intended to be used. So I am just intrigued about writing and AI’s.

I was reminded of what is perhaps the fundamental rule of technological progress: “if something can be done, it probably will be done, and possibly already has been”.

  • Edward Snowden, Permanent Record

Maybe, gone will be the days where you will have spelling mistakes and other typos. LLM’s will take care of that. Maybe we will be able to recognise human generated text the same way as we recognise spam now? Like the obvious spelling and grammar mistakes they had, AI generated articles might have something off about them. Now, state actors won’t be trying to put in the effort to care about any of these. It will be like grammarly on steroids. But it is also going to make it harder for us to differentiate AI generated articles and it’s agenda. The propaganda machines and corporate marketing team will have immense resource on persuading people around with very little resources. This means they could scale the operation to a whole new level. Focusing on minute areas which are ignored right now.

But from a reader’s perspective, the changes are little. But maybe more serious. It’s not just that we will be coming across many more AI articles than before. But we will also have to focus more on fact checking more than ever before. Something that we have been learning for a while now. So the end user’s responsibility of fact checking has been expanded from non-reputable sources to even reputed sources as well now. This is considering the rumours of companies and news media outlets laying off staffs and using AI to write articles. How do you know whether New York Times is using AI or not for their articles? I am not saying NYT only publishes correct things, but they are still reputable compared to many. I mean, how will you know if it is purely written by LLM’s or it is just LLM assisted?

So am just waiting to see how it would turn out in the end. When people doesn’t just use it when necessary but also by default and always. The effect it will have on the personality of a blog post. The emotions it’s supposed to share or convey. Especially if people using substack, medium and similar platforms starts using it. They are not your ordinary indie web folks who are maintaining their own personal website. This is more of a wonderment than a complain really. Will it make most of the writings similar? Or will using different LLM’s result in different styles? What would be left at the end?

By the time I finished the first draft for this article, Paul Graham actually tweeted about this with his prediction on the topic:

Paul Graham's tweet on AI assisted writing

Paul Graham's tweet on AI assisted writing

But during a Telegram discussion on the same topic, someone made an interesting point on AI/LLM usage effects while writing:

  1. For people who know how to write and how to think - yes, if they stop writing, they stop thinking
  2. For people who do not know how to write, but do know how to think - they have to write prompts and that’s where they exercise their thinking, while AI helps them to write
  3. For people who know how to write and not how to think - well, they will just use AI and get things done faster
  4. For people who do not know how to write and do not know how to think - then prompting does no harm to their thinking.

This is a very interesting set of points. This puts a lot of context unlike the mostly black and white opinions I have seen on the topic. But depending on the topic expertise, wouldn’t we be exercising all these kinds in different scenarios? I have often observed that the things that I already know is forgotten in a different/scenario topic. Information which I can re-use from programming for instance. As a programmer, I know that a big task can be fix more easily by making it into smaller tasks. But when an social/physical often presents itself where this can be easily utilised, I often forget this. Making me re-learn all over again. Only to realise that I have done it later while thinking through the scenario. We are often in self imposed containers when it comes to learning these things. We don’t often efficiently carry forward the learnings that we can in a different setting. We re-learn them.

As for my writing, I usually come back to most of my published blog posts. Make some edits and improvements. While not all, a lot of my posts are updated from time to time. The process helps me update the article regularly based on my evolving perspectives. Some of the parts of my website are intentionally manual. I enjoy the entire process of writing and publishing. This is because I do this whenever I feel like it. Keeping the novelty of it all in check.

Then there is reading blog posts as well. The consumption part. It always is inspiring to see other people’s writing. The way they write. The thread of an article and how they structure it. These will eventually result in memorable reads. Here are a couple of very different posts that made a huge impression on me:

I come back to them. The emotions speaks for itself in these articles. I wonder how the emotions will evolve with AI articles. Emotions probably would be patterns which the model might figure out without even knowing? But even if it provides the best output, aren’t the writer missing out on the whole process of writing? The joy/peace involved in it? Soon a whole new generation will be growing up which didn’t know another time where LLM’s and their assistance didn’t exist. Just like people don’t know an era without broadband, an era without dial-up internet and an era without internet at all. There will be a generation who know only LLM/AI assisted writing and life. Will we be eventually outsourcing our writing to machines? Just like how we stopped cursive handwriting? When I write in cursive, people say it is amazing. That my friend is because you haven’t seen proper cursive handwriting. Mine is not that great. On the other hand, my dad has one of the best and fluent cursive handwriting. Him on a bad day is mine on a good day.

I suppose, just like the re-emergence of fountain pens and it’s enthusiasts, there will be a portion of people on the internet who will still be writing on their own. We already have a indie web community who are trying to preserve the internet before it went commercial. Again, would that be great? Bad? I don’t know. But I started writing because I have something to share. I mean, we always will have something to share. But the clarity I get during the process of writing things makes all the difference. Not to mention, I ended up loving the entire process which all started because I had something to share - the publishing part. Maybe that is because I fall under 1st kind of people from the above list using this tech? In anyways, pouring my heart out and figuring it out has been humbling in a way. Let’s hope me pouring my heart out is enough going forward as well.

← Open Source projects and Non-profit 501(c)(3) vs Non-profit 501(c)(6) My memorable OpenStreetMap edits →