I was an eager user of Claude Pro, and I bought into it out of FOMO. Many people were building their work environments almost entirely on coding agents, and I wondered how I could innovate a new workflow for myself. I ended up using Claude constantly, but I didn’t realize how my use of it was affecting my mental health.
Before I go on, there are a few distinctions to draw. I am not a strict skeptic of LLM technology, but I have a more nuanced view. A strict skeptic are the people who swear off the technology for good, and I understand their viewpoint. I’m not that skeptical though. The technology is great and there are some spectacular use cases for it, but I find that the consistent application of coding agents to be concerning. I feel that when I use it I grow lazy and I stop investing in the work I’m doing.
On one occasion, I found that not caring for my work was a really big hurdle to overcome. I was driving to work, and when I arrived I dreaded getting out of the car to enter the office. At the time, I was using LLM technology heavily, and I found little interest in what I was doing. Since there was little interest, then I really didn’t want to go to work. This dread shocked me, and it took me a while to figure out that I wasn’t interested in my work at all because I wasn’t engaging with it.
So, I've been returning to the fundamentals of programming in an attempt to find engagement with programming. By working without LLM technology, I have found deep satisfaction in breaking down concepts and understanding how things work. In once such case, I prepped for a conference talk where I forced myself to learn Canvas API by just exploring it. Because I engaged in the struggle with learning, I felt much more interested in the API. For example, I am happy that I get to use trigonometry after 15 years of not having any reason to use it.
And so here I am. I am refining my use of LLM technology, and the uses I find are much narrower than what I thought before. I find that if I am stuck, then I can query an online chatbot on what I am missing. I don't let the chatbot do the whole work for me, because I feel that I wouldn't be helping my future self.
My mental health has improved significantly too. By removing LLM from my workflow, I feel more engaged with my work and it’s not something that I am dreading.
So I will dedicate to not giving out slop of any kind. I previously did use slop content, but I am not going to do that again. I want to be satisfied in my work and I want to encourage myself to be creative and natural. I have canceled Claude Pro, but I will use the free version to give feedback on what I make.
I hope that my quality of my work reflects my intentions, and I believe it will. When I do the hard work of understanding how things work, I know that myself and my work will be better. I look forward to challenging myself in the future on what things I can learn and build.