Struggling with Perfectionism in the Age of AI: My Experience with LLMs

Dear Invisible Friends,

As my second blog post (after re-launching my abandoned blog), I wanted to do a “raw” writing (I mean without LLMs) about a self-reflection.

🪞 Naming the Thought, Letting It Go

Maybe in a similar way that we apply all sorts of filters to our pictures (and of course make-up, duh!), I tend to do the same with text. I want to make sure it is readable, user-friendly (depending on my audience), and that it conveys the message.

However, by overusing Large Language Models (LLMs, more on that in future posts), I over-sanitize my text, being super perfectionistic, until maybe making them soulless. It may not matter in technical writing (like those I do for work), but it can also happen in emails. Of course, I always review the output of the improved LLM text to ensure it accurately expresses my opinion and message.

A person with long, blonde hair wearing large glasses and a choker necklace, standing on a train platform with railway tracks in the background on a sunny day.
Selfie to prove I’m not an AI generated person (the letters on the shirt, despite mirror image, are real)

🧠 Writing Without AI – Prehistoric Tricks

English is not my first language, and of course I want to avoid spelling or grammar mistakes. In the past (circa 2006), I used to Google to search my intended English expression between “” to check if it was a real sentence or a weird verbatim translation from Spanish, my mother tongue.

A little bit on “text mining” pre-Generative AI. I used “Control+F” a lot within a document, and just like magic, I could find (and copy-paste) the relevant information I needed for work or school.

Of course, now it’s super efficient to upload PDFs to the LLM of your choice and Turbo-“Control+F” the s***t out of your documents, but I did what I could back in the old days.

An elderly man sitting on a stump, telling a story to a group of children sitting on the grass, with trees and clouds in the background.

🧩 A Little Glance into My Thinking

I am thinking (and low-key planning) a blog post on the energy consumption of using LLMs and its impact to global warming and the planet. To keep myself in check, and to show that I still got it (soul, I mean), I decided to rawdog this post and avoid the use of LLMs, regardless how squeaky-clean (is that how we spell that, right?) it must have been by doing a quick LLM fix.

This is by no means a judgment to those reliant on LLMs to quick start writing a page. I know that the blank page is a dreadful place. Sometimes, I’ve been there too.

I also don’t have any professional or financial goals with this blog, it’s to express my mind, in this case, unfiltered.

Maybe it’s more my style to spend hours and hours writing, but I have things to do, so do you (I bet). My brain is telling my to do some catchy conclusion and ending and leaving a question to the audience, here I go:

So, I proved to myself that I can write both with and without LLMs, the quality of the output is not for me to judge. I think both have their advantages and disadvantages, and of course the choice to be “with or without LLMs” can be determined case by case.

A black and white image featuring the band U2, with the text 'With or Without You 1987-2023' prominently displayed above them.

🎨 Which Writing Style Do You Prefer?

And what is your opinion? Do you prefer to write “rawdog” or to embellish the text with LLMs? Leave your thoughts in the comments.

RoxenOut!


Discover more from Dina RoxenTool

Subscribe to get the latest posts sent to your email.

Posted in

6 responses to “Struggling with Perfectionism in the Age of AI: My Experience with LLMs”

  1. Cris Avatar

    I think I’m going to love this blog, especially when you write posts “rawdog”, straight out of your beutiful mind and heart. Stay unfiltered, Di! ❤

    Like

    1. dinaroxentool Avatar

      Thank you so much for reading this post and for your feedback!

      Like

  2. John Narváez Avatar
    John Narváez

    Cuando sospecho que estoy ante un texto escrito por un robot, pierdo el interés y dejo de leerlo. Supongo que un escritor robot solamente puede encantar a un semejante suyo.

    Like

    1. dinaroxentool Avatar

      Exactamente, los textos generados por LLMs aunque pulidos y correctos, pueden resultar sosos. La perfección gramatical va con el costo de la pérdida de la originalidad. Yo estoy buscando el balance y el punto intermedio, dependiendo del caso. Gracias por leer mi post.

      Liked by 1 person

  3. […] Struggling with Perfectionism in the Age of AI: My Experience with LLMs […]

    Like

  4. […] few weeks ago, I told you about my AI-free writing challenge and my struggles with perfectionism. Since then, I’ve shared my thoughts on AI and education, especially the panic around ChatGPT […]

    Like

Leave a reply to dinaroxentool Cancel reply