I’ve been writing the next entry in my “Reverse Engineering Crazy Taxi” series, and while doing so I read something I wrote in part 1 that made me physically cringe:

That’s because Jasper requires each game’s renderer to operate using just the game’s original files; no Blender exports, no pre-processing with modding tools, just rendering the game from its original files, and quickly enough to run in a browser.

To me, the underlined sentence’s construction of “no X, no Y, just Z” just smacks of AI generated text. But it isn’t! I remember exactly where I was when I wrote this with my humble meat-brain, and if not for LLMs overusing this phrasing, it’d be a fine sentence.

And yet some part of me actually feels compelled to change words I wrote myself lest they be incorrectly perceived as AI slop by a reader.

That’s because I just don’t value content generated by AI as highly as stuff written by a person, and it would be really disappointing if, after the hours and hours I’ve poured into this series of posts, a reader who disdains AI writing as much as I do saw this line and assumed the whole thing was AI dreck.

But instead self-censoring and changing an otherwise ok sentence, I decided to write this post making my position on AI clear: I don’t use generative AI to write my prose or my code. I virtually never use it in general, barring the occasionally helpful result in the otherwise obnoxious Google AI overview thing. Indeed, nothing on this site is created using generative AI, and if for some reason I ever do use AI for something, I’ll explicitly indicate that alongside the content.

I’ll elaborate on why I don’t use AI below, but for those who still value artisinal human-authored content in The Year Of Our Lord 2026, hopefully this is enough to call off your well-honed AI detectors while reading my posts. And if for some reason you want to read yet another person’s opinion on AI, read on.

The labor theory of blogging

As I said earlier, I don’t value stuff generated by AI as much as stuff created by a person. I think there’s a lot to unpack about why that is, and the reasoning feels quite subjective and nebulous for me, but here’s my first attempt at putting it into writing:

I value the human labor that goes into creating things, since knowing about that effort cultivates a sense of empathy with the creator, and allows me to relate to them at a deeper level. This goes for written works like blogs & articles, but also to pretty much any media or art. Seeing brushmarks on a painting allows me to envision the artist at work: making decisions about how to depict a lock of hair, or dwelling on some small detail of a cloud. I can imagine how I myself might approach those things, appreciate the directions the artist took instead, and the entire experience is enriched. This social relation between myself and a work’s creator is a huge factor in how I appreciate it.

But I cannot empathize with an LLM, and its “decisionmaking” stems from an arcane and unknowable mathematical process, so this connection is completely absent for me in AI generated work. This saps so much of my ability to enjoy the work that even the barest hint of generative AI can taint an entire project for me. People often criticize the products of generative AI as “soulless” or “lifeless”, and for my money, that missing “soul” stems from the lack of an empathetic, social relationship to a human author.

And on top of these subjective factors, I haven’t even gotten into the ethical and political problems surrounding the AI industry’s cancerous growth, or how AI generated code shatters the social contract for FOSS projects and turns my day job into a total nightmare! But these are things others have written about much more eloquently than I could, so I’ll leave it here for now.

What are we even doing here?

With all this said, I think it’s worth recentering a bit about why I even started writing on this site.

Nearly all the writing I’ve done here relates to the journey of figuring stuff out. To me, that’s an incredibly important human experience, and cultivating curiosity and problem-solving skills is why I pursue topics like reverse engineering and mathematics in the first place.

Using AI to reverse engineer something, or to write a blog post about how I reverse engineered something, feels like it defeats that purpose. Sure, at the end of the day I’ll end up with a heap of words and code, but I won’t have had the experience of engaging with the problem or the ups and downs of experimentation. I also suspect I’ll have enjoyed the process significantly less than if I just worked at it myself.