Why I Use AI to Write

We've all heard it: "AI slop." And I have to say that a lot of AI writing is slop. It really rubs me the wrong way when I'm reading something that I know is AI, but it tries to be personal or gives a personal opinion. What really bothers me is when I read something that is obviously AI, and it is signed by a person, like they wrote it.

But does it have to be slop? If I'm reading a post because it has good information and is succinct, I don't mind if it is AI generated. Good information is good information. AI is actually very good at creating technical content; it tends to avoid the usual AI "tics". Things like "this is the thing that everyone ...", or adding filler words where they don't sound natural. Given the correct instructions, it can be very useful. I have used AI, and will continue to use it, for purely technical writing. I could write the posts, but to be honest, I think AI does a better job, and in much less time. I can "create" a lot more content with help from AI.

Can't someone just go to ChatGPT or Claude and create the content themselves? Yes, they can, but I try to provide content that might not be obvious.

Here is an example of a well-written blog post, written by AI: Ohm's Law in 5 Minutes. I did a bit of editing and pointed out two AI tics; Claude then cleaned up two more. With a little extra work, AI can do a nice job.

Here are some things I do to improve my AI content:

• Give it my voice. I gave Claude three essays that I wrote in college and two blog posts that I recently wrote. I asked it to create a file that summarized my voice. Now when I ask Claude to write an article, I tell it to use my voice file.
• Tell it what not to do. As part of the instructions, add "avoid common AI tics." When the article is done, I clear out the obvious AI tells anyway.
• Take the document to another LLM. This is very powerful, and I wrote an article explaining it in detail.
• Ask for references. Tell the AI to provide references for every factual statement. This helps avoid hallucinations, but you will still have to verify them, or have another LLM verify them.
• Make sure I agree with it. Before anything goes out, I make sure I agree with the content.

Why do I bother with changing the voice? I'm not trying to hide the fact that it was written by AI. I just want the post to have the same feel/vibe as my writing. I guess I like the sound of my own voice (what man doesn't, LOL). I really just want it to sound more natural; AI tends to move towards its "center," which rarely sounds like a real person. I'm forcing it away from the center and towards a real person.

I have made a stand: if I use AI to write an article, I will let everyone know at the end that it was generated by AI and edited by me. If I wrote it myself and used AI to help with spelling and grammar, I will say that at the bottom too. If you see a post that says "I, George Clay, wrote ...", you will know I wrote it. I haven't done that on some of my past posts, but I will going forward. I want to be completely clear about who wrote what.

I'm not ashamed to use AI; it is a tool. I use an impact driver to drive in screws; that doesn't make me less of a man than using a screwdriver. If we are going to use AI, let's be honest about it. If you are trying to hide it, then you are telling us something about yourself.

Let me know if you think I'm crazy. (OK, I am, but that is beside the point.)

I, George Clay, wrote this post in whole. AI was used to clean up my grammar/spelling, and point out any gaps in continuity.

P.S. This wasn't part of my original post, but I had to add it. After writing it, I asked Claude to look for spelling/grammar/continuity errors. It pointed out a few spelling and grammar errors, which I fixed. Then it wanted to replace "tends to avoid" with "it avoids most of them but a few always slip through," which sounded to me like injecting an AI tic into what I thought was a perfectly understandable sentence. Don't always believe what AI tells you!