If you use AI seriously, you already know how good it is. And if I’m being honest, the latest version of Claude feels like I was taken through a wormhole. Yes, it’s that good, but part of me is afraid that the new frontier it dropped me into might make me something I regret.

For the longest time (past three years), I have looked at Gen AI as largely a toy, writing pithy songs about Volkswagens to tunes of John Denver, and other high-value prompting sessions. 🙂

But increasingly, I’ve noticed a bit of a shift happening, that is, the mental model of opening a Claude chat and then, well, offshoring my thinking and tasks. Don’t get me wrong, I have marveled at the output and efficiency, but I do think there could be something in me that is slowly being overwritten. Mental drift, if you will.

If You Can’t Design on Paper

When I was in college, digital tools like Adobe Illustrator had just come out, and as students, we wanted to use them as much as possible.

“If you can’t design on paper, you can’t design on a computer.”

I can remember my design professor Frank Cheatham saying that statement has always stuck with me. The underlying notion is that skills are developed through mental work, iterations, and refinement. If I forgo any of the steps of the design process for tooling, then the outcome will be reflective of that.

Sure, the lines in a logo might be crisp, but the idea will be weak, and therefore, the end result will also be.

Tooling vs. Thinking

We are in the age of massive (no hyperbole) acceleration of tooling. Tooling that can kick out N number of options on demand.

But how good is the idea that went into it? Did I go through the mental exercise of thinking about the audience, what resonates, what I’m trying to communicate, crafting a solution, and then judging things based on that criteria?

Now I know that good processes can also be replicated within a tool, and I’m not advocating that you don’t use them, but I have personally witnessed myself skipping the process due to the ease of just prompting and then not really judging the output in the same way.

I am, by nature, a Luddite, but the lure of Claude has become quite strong. Especially Opus 4.6 (yes, I know naming models will date this post quite quickly), but the fact remains that it is an impressive achievement worth mentioning specifically.

I just want to make sure that the tooling doesn’t replace my thinking.

In short, I want to use tools like Claude in a way that makes me better, instead of lazy, thereby adopting its name in the alternate spelling: clod /klŏd/

P.S. No animals or AIs were used/harmed in the production of this post.