Is this blog just more AI slop?

I tried reading Hypnocracy, which was written with an LLM. The book hallucinates about the “Berlin experiment” at the beginning, but never explains what it is. I looked it up, and this experiment doesn't appear to exist. I gave up after the first chapter because the book reads like a prompt response.

Seeing how bad it can get, can I use AI to write this blog? It's tempting to use an LLM to instantly produce polished content. And while I'm using AI tools in my writing process, I don't want my blog to resemble a collection of AI-generated posts.

Why I started writing

I started this blog because AI tools eliminated the psychological and linguistic barriers that had prevented me from writing. English is not my first language, and I wanted my blog to be accessible to a wider audience. I also wanted the sentences to sound grammatically correct and be free of typos. To achieve this, I use DeepL Write, an AI tool that improves phrasing and suggests alternative sentences. AI tools have given me the confidence to express myself clearly, refine my writing, and publish my work.

Rubber duck

Before writing, I use an LLM as a “rubber duck”. I send it my half-formed thoughts. Then, I review the output and actively reject anything that doesn't resonate. I repeat this process until everything is clear and structured. This helps me refine my thoughts. For example, I started my first blog post by listing new tools I had started using and enjoying. The back-and-forth with the LLM helped me articulate why I liked these tools. This transformed the post from a list of tools to an argument about sensible defaults.

Cold review

Once a draft is ready, I use LLMs as opinionated readers to quickly receive feedback on my writing and identify blind spots. I do “cold reviews”: I submit my draft with no context, so the model reads the article as if it were the first time. LLMs are good at evaluating tone consistency, flow, and the strength of my opening and ending. For example, this was my first ending for this blog post:

«AI tools are just a way to reduce friction. I see my blog as a thinking exercise. I strive to convey my ideas as clearly as possible, holding myself to high-quality standards.»

And the LLM's comment about it:

The closing two sentences ("AI tools are just a way to reduce friction" and "I strive to convey my ideas as clearly as possible") are both strong but feel slightly disconnected from each other. Tying them together more explicitly would give you a stronger ending.

I am responsible for the quality of this blog. AI tools should reduce friction, not lower standards.