Hey there, I’m Eeshan. I write about local AI, data science, and how to use AI, with a bias towards human-centric approaches to working with new technologies.
You can find some of my interesting side projects at eeshans.com.
Once upon a time, in the age of Eminem & Avril Lavigne
If you’ve been a reader of my Substack since I started it about 6 months ago, you have probably noticed that I write about AI a lot. I’ll be honest, when I started it, I was kind of chasing the hype a bit, and was truly in an experimentation phase to see what’s possible or not with the new shiny tools we got to play with. As time has gone by, I’ve learned a LOT more about how to truly use LLMs, in specific, narrow applications that have net positive value, and ideally with local models so I’m being energy efficient and privacy-focused as well. So I’ll keep doing that since I think it’s the right long-term play once the hype & bubble have died away.
I also heavily used an LLM to write my first post, for which I kinda still feel “guilty”, and you’ll understand why by the time you get to the end of this post. But now, I’ve completely stopped using AI for any writing at all. I never really used it for professional writing since the stakes are always too high, and I’m quite good at it professional writing anyway, so now I’m just trying to phase it out completely because I can’t bear the guilt anymore.
I have been grateful to have lived through the golden age of the Internet, and the early 2000s in general, where there was just enough technology to have had a net incremental effect on creativity and innovation, while still allowing us to maintain the real, in-person connections we had with people. I won’t get all nostalgic here and talk about how everything was so cool in the early 2000s. I’m of course going to be biased because most people usually love their childhood / teenage era and I am no different. We had just enough Google, just enough Facebook, the greatest movies ever (LOTR trilogy, Harry Potter, Dark Knight, Gladiator, V for Vendetta, Up, Wall-E, Kill Bill and many more), and prime Eminem & Avril Lavigne.
On writing
But in this post, I want to talk about a very specific thing that I’m not just nostalgic about, but in a bit of mourning to have slowly let one of the skills that make humans truly human degrade. It’s about writing.
This thought came to me when I recently had a very thoughtful conversation with fellow substacker / substackian , where he was generous enough to compliment me on my writing style on this Substack. To which I was of course flattered, but also thought that this is nowhere near my best work. I’ve been journaling since I was 11 years old, and have also had a blog since 2008. This might surprise my readers but I used to churn out prose and poetry with ease in my teens and tweens. I credit my Mom for inculcating this wonderful skill & hobby into me.
But that old blog just sparked so many memories and I realized that I had made the blog private while I was thinking about migrating the contents to Ghost or Substack, and just never completed any of it. The blog was just hidden in purgatory outside the eyes of any of my 94 subscribers.
So I decided to revive it and make it public again. I spent an hour going through almost all 132 posts, and it just was so bittersweet for me.
Sweet of course because of the nostalgia and rumination about the time when I wrote any of those posts, poems, stories. It wasn’t that I admired my writing style, which was always imperfect, but it used to flow so naturally out of my chaotic, introverted mind.
Bitter, because I felt bad that I don’t do this anymore (besides the more technical posts on this Substack), and also because of this “guilt” that how did this guy allow LLMs to write on his behalf for even a post or two. It was blasphemy, it was sacrilege.
The blog from the old Internet
Here’s a link to the old blog: Eeshan’s Blog – An Introvert’s reflections..
You have been warned. The content in that blog may seem too imperfect, too illogical and too human. You might not be used to it anymore.
It’s also no surprise that the first post I ever published on this blog was about writing itself. I’ll just let 2008 Eeshan communicate the gist of this post to you, because I don’t think 2026 Eeshan can write like this anymore.
And it was never about having writing skills at the level of Mark Twain or William Wordsworth, it was always about the sheer imperfection and purity of it all.
Writing is about self-expression. It’s about ownership, that you mean what you say. And that there’s a real living breathing individual standing behind those words.
You know in the old days, a promise was made when one person said to another, “I give you my word”. That was a verbal contract of trust, and you would only believe the person if they had a strong history of actually “keeping their word”.
Today, we can’t say any of this. In so many places, especially online, a person’s word is not their own; it’s an approximation of the aggregate words of all the past owners of the training data embedded within the weights of these LLMs. It’s just empty calories, sentences and paragraphs put together mechanically through the rules of grammar. AI-written text has no owner, no real meaning, and no soul.
Why I think people use AI to write, and how to tell it apart
In some twisted way, I think that writing with LLMs is a sloppy cure for laziness and incompetence. It’s why AI-written text is called slop, because it really is sloppy in the traditional sense of the word.
It’s a cure for laziness, for those who might have decent writing skills, but don’t want to put in the time to write, OR churn out “content” for other purposes.
It’s a cure for incompetence, for those who just don’t have any decent writing skills (for whatever reason), and AI-written text for them could actually be far better than what they might have written themselves.
What all this means is that so much more written text now exists in the world that would not have existed if LLMs weren’t available. I work in data science, so I often think in counterfactuals or “what would have happened if X didn’t exist” (not talking about X / Twitter, but that is a worthwhile thinking exercise). If LLMs didn’t exist, we probably wouldn’t be getting shoved with all this in our faces. It’s wild when you think about it that way.
How do you tell if something is AI-written? There are some great resources here and here, written by people who’ve done far more research on this topic than I have, but below are some of the easy ones to spot:
Bullet points and nested lists (the most obvious one)
2nd and 3rd person only, never in 1st person
Constant parallelism - it’s not about X, it’s about Y
The em dashes (—) - this one is sad, because there are genuine uses for these, but now real people get called out if they use em dashes
Hollowed out, generic statements (you need to pay attention to this one)
The worst thing about this is that we’ve gone from GPT-2 which was too dangerous to release in 2019, to Claude Fable & Mythos that are too dangerous to release in 2026, and all this slop writing is still exactly the same. I’ll just leave it at that.
If it wasn’t obvious enough, reading with AI is just as bad
Reading / summarizing with AI is just as bad. It’s very similarly a cure for laziness and incompetence. This is an extreme example, but you can’t say you “read” Crime and Punishment by Dostoevsky, by reading an “AI summary”. You just can’t say that. Dostoevsky, in particular, is one of those legendary authors whose words hit you DEEP inside your soul. If you’re reading a “summary”, then it was probably a net negative learning experience. You felt nothing from the words of the greats, and you claim the contrary.
If you ever encounter a situation where you need to have an intellectual debate with another person, but their knowledge is based on AI-summaries (also Google AI mode now), then I leave you with this classic “not you” LeBron meme to share as a response. This is an all-purpose, generational meme. Thanks LeBron!
Lucro 💫 on X: “LeBron Not you already a classic 😭 https://t.co/aYy0cHsp9q” / X
Any form of human self-expression is worth preserving
Reading, writing, drawing, painting, dancing, singing, or any activity where the output is a representation of your own self is totally worth preserving. For me, martial arts has always been one of those things. It’s an expression through the body and through actions, less through words. Perfect for that introvert / extrovert mix personality type.
The new & powerful content generators will continue to exist to fuel consumerism, and there’s not much we can do about it. But my simple advice would be to just keep your creative hobbies alive. If you don’t have any, then it’s worth picking one up no matter what stage of life you’re in. You can get better too. And how do you get better at any of this ... by doing the thing, over and over again. It’s as simple as that. No magic trick, and certainly no shortcut.
I’m going to apply what I preach, and keep my writing authentic and AI-free. I hope my readers will appreciate that. It will be clunky, but it will be very human.
I will of course still write about “AI”, because I’m a technologist, and my goal is to continue figuring out human-centric approaches to leveraging AI as time goes by, especially once the hype subsides.
If you got this far, hope you enjoyed this little piece. Please drop a comment or a note to let me know your thoughts on this.