AI Experiment

The first iteration of the site was a photography portfolio. Then a travel blog. Next it was mainly a feed of links to my journalism articles.

I was never satisfied with the content management systems I paid for. The common denominator was that publishing content on the site was slow, with too many steps.

Today, the site is a landing page for my data infrastructure work. The current version doesn't use a third-party content management system (CMS). Nor is it built on a template that I selected after scrolling through dozens of options.

Instead, generative AI helped me build the site. It produced the code and drafted some of the landing page and marketing copy based on my prompts. The initial version was done in two hours and at no additional cost. I canceled my CMS subscription. AI allowed me to revamp my personal spot on the web while staying more focused on executing on client work and spending time with my family.

Iterating on the code with AI and working out the hosting and a publishing workflow was a useful learning process. After a few weeks of use, the process of publishing content remains much faster than using a traditional CMS.

At the same time, I'm more concerned than ever about being so dependent on AI that my own thoughts and work are effectively outsourced.

AI Limits

There are two areas where I am pulling back on AI use.

The first is blogging. I see a distinction between website copy, which is functional, and blogging, which is my thinking. Writing helps me distill thoughts and find focus. I experimented with what AI could produce when I gave it an outline and key ideas. I felt like the reader was being misled, since I wasn't the one putting the words together. The outputs were generic too.

Going forward, AI won't be used for blog posts on justinvela.com. It won't be used in the brainstorming process either. At most it will be used for proofreading since I don't have regular access to a human editor.

The second is imagery. I will not use AI for any imagery on this website, still or video, except for basic background gradients and textures. All of us should be asking what's real and what's not when looking at a screen. There's plenty of AI imagery around.

I won't be contributing to it.