AI is everywhere right now, and you cannot escape it. The tech bubble is full of articles and think pieces that either promise a brave new world of software engineering or the downfall of the craft. And let’s not talk about all the companies that are shoving AI features in the faces of their customers if they want them or not. I am not even the first one to comment on this. Rants about AI are now as ubiquitous as the technology itself. AI is everywhere, and this is why I want to talk about why I don’t use it to write this blog.
I think I need to get a bit more concrete here. I’m not against AI per se. In my current job as a Data and AI Consultant, I work with it every day. I ask it technical questions when search engines did not turn up any usable results1. I increasingly use coding agents to do routine work1. Yes, I even used a coding agent to fix social card and Open Graph metadata generation on this blog because I couldn’t be bothered2. I am not here to make any moral judgment about AI, whether you should use it at all or not. This post is simply about why the text you are reading right now and none of the other articles on this blog were created with AI assistance, and why I think it is important.
Using AI is all about efficiency. You can do more in less time. If you have a look at the number of posts on this blog and their timestamps, you can see that it was never about efficiency for me. This blog exists because I like to write about technical topics. I like to ponder about the exact phrasing of my words, not only the high-level structure. Sometimes, I have to stop myself to go over the same phrase over and over again because it doesn’t sit right with me. Does that mean my articles are rhetorically flawless? By far! But they are mine – my words, my thoughts, exactly as I thought them3.
Writing also helps me structure my thoughts. Normally, I do all the coding first, create an outline for the article second, and then write it. I could call it a day after the second step and just give the outline to an LLM. But more often than not, I find an issue with the outline or even the code while typing up the article. I’m not really sure if I would notice these issues by reading over LLM-generated text. The topic at hand becomes clearer with each paragraph I write. Even with this text right now, I notice how typing them out forces me to articulate my vague feelings about using AI for blog writing. Maybe I even would have changed my mind if the article became much too short.
When it comes to using AI for coding, I’ll probably have to go with the times, even though I like handcrafting code as much as I like writing. For the sake of efficiency (and keeping my job), I will use it and try to find joy in coding with it. Writing this blog will remain my refuge from progress, if you will. Typing out one post at a time to order my thoughts. Just for fun. Just for me. And now that you’ve finished reading, just for you.