Uncritical thinking

Two of my favorite writers took on the same topic from different directions this morning. The first to hit was Matthew Inman, who released a (very) long cartoon digging into the artist's relationship with the collection of technologies we call "AI." It starts with his observation that "even if you don't work in the arts, you have to admit you fee it too — that disappointment when you find out something is AI-generated." (Since it's a web comic, you'll just have to read it to get his full essay.)

Author John Scalzi also had some thoughts about AI, especially the volume of AI slop that consumes more and more of our attention online:

I think there’s a long conversation to be had about at what point the use of software means that something is less about the human creation and more about the machine generation, where someone scratching words onto paper with a fountain pen is on one end of that line, and someone dropping a short prompt into an LLM is on the other, and I strongly suspect that point is a technological moving target, and is probably not on a single axis. That said, for Whatever, I’m pretty satisfied that what we do here is significantly human-forward. The Internet may yet be inundated with “AI” slop, but Whatever is and will remain a small island of human activity.

The same is true for The Daily Parker.

I mentioned a few weeks ago that I've started working with LLMs as well. I have now used GitHub Copilot models Chat GPT 4.1, Chat GPT 5, and Claude Sonnet 4 to fix several bugs that have frustrated me for months. And to both Inman's and Scalzi's points, the LLMs help because I'm already a seasoned professional, and this just puts a couple more tools in my belt.

But as Inman points out, the AI slop we see today looks great to people who don't have skills. In my case, the bug fixes and performance optimizations that the LLMs suggested didn't work right out of the chat window, and I had to ask the models several follow-up questions before I got to working code. Even then, I had to carefully fit the models' outputs into my existing style and architecture, which on more than one occasion required taking a model's idea and doing something completely different with it.

So yes, keep using AI-driven productivity tools. Just don't call it art, and don't call it coding.