AI Policy

I've never had a formal AI policy for this blog, so I wanted to clearly define my thoughts on AI slop and how I'll use generative artificial intelligence on this blog.

Blog AI Policies

All words in a post are written by a human

I physically typed everything in the body of my blog posts unless explicitly mentioned.

For example, something on this site that was previously AI generated - the JSON-LD block on the Recipe SEO investigation was generated by GPT from an export of the recipe. The post on AI-generated home-server documentation includes explicitly AI-generated documentation.

All ideas are mine

I don't want to regurgitate an AI conversation or have AI give me the concept of posts to write. This blog's entire purpose is to have a space on the public internet to broadcast my personal thoughts.

You don't care about my Claude conversations. AI psychosis and the experience of using these chatbots make people confidently spew incorrect things because a slopbot confirms their existing opinions.

Posts are reviewed for spelling and structure by AI

I have frontier models review my posts for spelling and structural issues before I send them to my human editor. (Hi Mom!) I don't let the slopbot change items directly - I physically type those changes into Obsidian after using my human meatbrain to decide if the suggestions are worth it.

How I Use AI

I use AI at work

I'm about to draft a post about this - but in my role as an automation developer working in the Microsoft Power Platform, I am using Claude Code every day. I probably haven't written any Power Fx myself so far this month. I certainly am moving faster because of it, but my feeling of ownership over my work and my job satisfaction has significantly decreased.

I vibecode things at home

I'm currently using leftover Claude tokens to vibecode a podcast app for myself. I've played with open-source Chinese models - both self-hosted and cloud-hosted using OpenRouter. A personal favorite coding model for my personal stuff is GLM 5.3-Flash.

I also one-shot a migration from MkDocs Material to Zensical¹ for this blog using Opus 5.5.

Thoughts

AI makes me feel icky

AI art looks gross. Generative AI is trained on massive intellectual property theft - megacorps yoinked the entire internet without compensating the creators who make the content that makes the models useful. The massive energy costs from the compute required for these beefy models harm the environment. Data centers are being built everywhere in communities that don't want them, causing strain on the already-fragile electric grid.

Thinking about AI is exhausting and exciting

These opinions are not unique; I still thought it would be valuable for me to put these thoughts out there. AI seems to constantly be the topic of conversation, sucking all the air out of the room. It's exhausting and exhilarating at the same time. New capabilities are constantly coming out. I vacillate between excitement and dread. As a tech nerd, I still believe these models can be very cool. As someone who enjoys my career making software, I don't want to live in a world where I just ask a robot to do my role for me.