skip to content
san.is
Table of Contents

Let me address the elephant in the room—or rather, the large language model in the terminal.

Yes, I use AI to help write these articles. There, I said it. You can close the tab now if you’d like. I’ll wait.

Still here? Good. Let’s have a proper conversation about this.

The “AI Slop” Problem

The internet is currently drowning in what’s been aptly termed “AI slop”—that peculiar genre of content where someone types “write me a blog post about productivity” and publishes whatever emerges, typos and hallucinations included. You know the type: vaguely authoritative, suspiciously comprehensive, and about as nutritious as a supermarket sandwich that’s been sitting in the reduced section since Tuesday.

It’s the digital equivalent of someone asking a stranger to take their driving test for them. Sure, you might pass, but you still can’t actually drive.

The Irony Is Not Lost on Me

Here I am, a solution architect who writes about technical topics, using an AI assistant to help structure and articulate my thoughts. The irony is thick enough to spread on toast.

But here’s where I’d like to make a distinction that I think matters: there’s a rather significant gap between “AI wrote this” and “AI helped me write this.”

The Difference That Matters

Every technical article on this site comes from somewhere quite specific: an actual problem I’ve solved, a pattern I’ve discovered through trial and error (heavy on the error), or a hard-won insight from years of staring at integration logs at 2 AM wondering where my life went wrong.

The AI doesn’t know that HubSpot’s rate limiting documentation is optimistic at best. It hasn’t debugged a middleware that worked perfectly in staging and then spectacularly failed in production because someone forgot about timezone handling. It hasn’t had that particular sinking feeling when a client says “we need this live by Monday” on a Friday afternoon.

I have. Repeatedly.

What the AI does is help me take that messy constellation of experience and turn it into something coherent. It’s less ghostwriter, more editor with an inexhaustible supply of patience and zero judgement about my excessive use of parenthetical asides.

The Editorial Process (Such As It Is)

Here’s roughly how this works:

  1. I have a problem. I solve it. I think “someone else might find this useful.”
  2. I explain the problem and solution to an AI, along with all the context and caveats.
  3. It produces a draft that’s technically accurate but often reads like a Wikipedia article had a child with a corporate memo.
  4. I rewrite the bits that sound like they were written by someone who’s never experienced frustration.
  5. I add the jokes. The AI’s attempts at humour are… let’s say “enthusiastic but misguided.”
  6. I verify everything still makes technical sense.
  7. I decide whether it’s something I’d actually want to read.

If the answer to that last question is “no,” it doesn’t get published. My name’s on it, after all.

The Value Question

Here’s what I keep coming back to: does the use of AI in the creation process make the content less valuable?

If someone reads my article about API rate limiting and it helps them avoid the mistakes I made, does it matter that I had help organising my thoughts? If a junior developer reads about middleware patterns and learns something useful, should they care that I didn’t personally type every semicolon?

I’d argue the value of content lies in three things:

  1. Is it accurate? (I check this. Thoroughly. The AI hallucinates sometimes.)
  2. Is it useful? (It comes from real problems I’ve actually solved.)
  3. Is it honest? (Well, you’re reading this article, aren’t you?)

What I Won’t Do

I won’t pretend to have expertise I don’t have. The AI can write confidently about quantum computing or brain surgery, but I can’t verify any of it, so I won’t publish it.

I won’t pass off generic AI output as insight. If I haven’t actually done the thing, I’m not going to write about it.

I won’t stop learning. Using AI as a crutch rather than a tool is a fast track to becoming obsolete. I still need to understand everything I publish, or I can’t answer questions about it—and worse, I can’t build on it.

The Uncomfortable Truth

The uncomfortable truth is that the “authenticity” we prize in writing has always been somewhat performative. Every writer edits. Every writer has influences. Every writer has had a moment where they thought “I should probably make this sound less like I wrote it at 3 AM after too much coffee.”

AI just makes that editing process faster and more accessible.

The real question isn’t whether AI was involved—it’s whether the person publishing the content actually understands it, stands behind it, and has something genuine to contribute.

An Invitation

If you’ve found any article on this site useful, I’d argue that’s worth more than purist notions about how content “should” be created.

If you’ve found errors, inconsistencies, or things that don’t match your experience, please do let me know. That’s how we all get better—humans and AIs alike.

And if you’re still uncomfortable with AI-assisted content, that’s entirely fair. There are plenty of excellent writers who do everything the traditional way. Their work is valuable too.

As for me, I’ll be over here, solving integration problems and writing about them with whatever tools help me communicate most effectively.

The AI typed most of these words. But the mistakes I’ve made, the lessons I’ve learned, and the decision to share them?

Those are still entirely mine.

This article was written with AI assistance. Obviously. That was rather the point.