Fuck AI Detectors.

I use AI quite a bit, whether it’s to help me plan things, answer questions, generate images to preview things, or even tell it personal things. I’m not the only one who does this.

I never use the images I create with AI as a final product, not because they’re bad (AI-generated things aren’t inherently bad) but because this is a human and humanist website, and because, although the fact that something is human doesn’t automatically mean it’s more valuable, I prefer imperfection in cases where it’s not a bad thing.

This is a completely human space.

My character is a robot, but not an AI. It’s a fantastical being, theoretically a human with an artificial body.

You might be wondering, “What does this have to do with anything?” Well, I discovered that there are online tools to check if a text or image was generated by AI.

And I’m quite concerned about the result I got for an image of Mastedore the robot, drawn by Ecl1p23:

Can you believe this shit?

How dare this piece of dogshit mark this drawing as 71% AI? An image handmade by a human being who dedicated valuable time to my idea and liked it? It offends me.

I have proof and timelapses of my friend making this drawing from scratch. I have nothing to hide about the artwork of Mastedore the robot because I have everything here, the most human process possible in its creation.

And this fuck doesn’t end here, the text is even worse.

Several months ago, I made my first post on this blog, a short text but in my style:

I don’t have a screenshot of this, but I had pasted that same blog post I just mentioned into two of those shitty tools (because there are tons of them), and both said that the text was ~80% AI. What???

I swear on my mother’s grave that I wrote that text entirely on my own (just like all the other posts on my blog). It seems absurd and ironic that my own text was flagged as 80% AI, and then I pasted a response directly copied from Claude and now it says 30% HAHAHAHAHA.

That’s when I realized how dangerous these dogshits are, because if they’re used to determine someone’s academic or professional qualifications, that stuff can screw anyone over without remorse.

The other issue is that you can simply create a skill for the LLM to do the exact opposite of what the guides say about detecting AI text. LOL. That thing is easier to fool than to use properly.

Many say that artificial intelligences are shit, but its detectors are even worse.

If you’re going to use an AI detector, please don’t use it to draw a conclusion about something, but rather to extract data based on statistics and probability.