“A computer is like a violin. You can imagine a novice trying first a phonograph and then a violin. The latter, he says, sounds terrible. That is the argument we have heard from our humanists and most of our computer scientists. Computer programs are good, they say, for particular purposes, but they aren't flexible. Neither is a violin — or a typewriter — until you learn how to use it.”
— Marvin Minsky

The One About AI

I’ve always taken pride in the process of growing as a creative. Whether that manifests in how I do things in my personal life (learning new recipes, learning new crafts, making art), or as a professional (improving my systems thinking, improving on a tool, discovering a shortcut), a lot of meaning for me has derived in having an exploratory mindset. I tend to avoid shortcuts, because my creativity and efficiency are rooted in a hard-earned understanding of the fundamentals. So my relationship with AI has been more as a tool to explore a lot of ideas or to do deeper dives. My main usage of the tech is chatGPT, and though I have used some tools touted to be AI, I have often felt that there is something lacking in the execution that I have seen so far of it.

I like the quickness of being able to throw out a lot of ideas and the playfulness of “talking through” where my thoughts lie with AI but I worry about some things. Namely:

It often plays a game of “Two Truths and a Lie.” Unless I am really well versed in a subject enough to see the inaccuracies, I prefer to start with AI and reinforce it with research.
There sometimes is an uncanny valley feeling in the output. For example, a UI might look like one would expect, but some ineffable details feel off.
For ideation it’s important to keep in mind that one’s chats can very very easily become an echo chamber that is more dangerous than a skewed algorithmic feed. Sometimes I have to adamantly demand for the chat to ensure it calls out my own inaccuracies instead of potentially blowing up my ego.
By bypassing processes, it can become difficult in time to articulate the foundation of what one is doing.

When I decided to build my site and blog, I used AI to talk about my main idea, theme I wanted to go for, and had it help me organize an ad hoc timeline that made sense for schedule and needs. I discussed the visual layer to help myself wrap my head around what I wanted to make. I had it walk through what tech made sense for my goals (I built this in hugo and use sass for the visual layer) and discussed the workflow and pipeline. I did not want it to write my content, design the UI for me, or write anything beyond a few code snippets. It helped me troubleshoot, validate, and gave me the confidence to have this site up in a small window of time. I believe in AI as a tool but I don’t think I would enjoy my life as much if I didn’t have the friction to learn things in depth.