A running list of thoughts from using LLMs. Note that I started writing this in early 2026. Maybe in a few years much of these thoughts will be oudated or proven wrong. Also, much of what I’ve learned about myself and people from using LLMs overlap a lot with lessons on how to use LLMs effectively. That might be an interesting read as well.

How LLMs think

LLMs “reason” by producing a “chain of thought”. If you look into how ChatGPT or Claude is “thinking”, it’s really just producing a bunch of text, going through the problem step-by-step, with words. “First, I do this. Then, I do that. Wait, let me double check…”. And somehow, by doing this, they are able to solve some very complex problems, and even create (arguably) new knowledge.

So just from training over the entirety of human writing, LLMs gain the innate ability to reason and become “intelligent”. Interesting possible conclusion from this: writing is thinking. Whenever you are writing, you are producing your own “chain of thought”. And so when you delegate writing to LLMs, you are delegating your thinking as well.

I realize when most people use LLMs to do their writing, they’re doing it for automated engagement on social media or to get their homework done with faster. But I do find it unfortunate, because writing is supposed to be an expression of your self and your thoughts. To delegate that to a next-token prediction machine is almost an insult to your own human intelligence and individuality.

I’d compare it to a guy who’s being invited to go hike a mountain by his friends, but refuses cuz he wants to scroll Instagram reels instead. Like yes Instagram reels is easier, but you’re missing out on something here. Some chance at meaningful human experiences and thoughts.

Consequently, I really dislike reading AI generated writing. I read and write because I want to see and convey individual thought, experiences, and intent. LLMs don’t have those. They have no individuality¹. And I really do consider it almost a personal insult whenever a student in a class I’m tutoring for submits writing so blatantly AI. I don’t mind using AI to code, that’s an engineering task. But every once in a while, the assignment invites you to make whatever visualization you want, interpret the data from your own perspective, ask questions you’re curious about. That’s a creative task, and it’s an insult to yourself and human individuality to delegate that to an LLM. It’s also an insult to me cuz like, man if you’re gonna use AI to do your whole homework at least don’t make it blatantly so.

Agency

When working on a project, a lot of people don’t really know what to do without someone telling them what to do. I don’t mean this in a knowledge sense; I don’t know a lot of things when I’m starting a new project either. I mean this in a self-agency sense; when I don’t know, I find out. A lot of people I work with seem to not know how to find out, or maybe they just don’t want to bother.

This is probably very arrogant and overconfident for me to say, but because of this it’s often easier to have no teammates and just work with AI agents. If in both cases I need to give specific instructions, then at the very least with the latter it can get it done in 5 minutes with no onboarding needed, whereas with the former I need to explain to them all the background knowledge first and then they’re probably going to end up prompting their own agent anyway.

Being “high-agency” was a buzzword in tech twitter in 2025, but I really do believe in its importance, especially as AI gets better and better. AI is almost always going to be better at following instructions, so one should get better at giving instructions.