Xavier H.M. - Notebook

I’ve been using AI a lot more recently, and this questionnaire provided a good excuse to reflect on my changing habits and views surrounding the topic.

1. How was your first experience with AI models?

The very, very, very first time I used any sort of “AI model” was at work several years ago. I worked in fast food and we were having a slow winter night. I fucked around on some character AI site just out of curiosity. One of my coworkers proceeded to come over and engage in a hypothetical with a storytelling AI chatbot wherein he was president of the United States a la Saints Row 4. It eventually devolved into some crazy helicopter action scene where he was gunning down the American populace or something.

The first time I used AI for myself was ChatGPT. I think initially trouble shooting something on my computer. As I got more into webdev I would ask questions about CSS/HTML rules that I couldn’t remember.

2. Do you use AI or are you completely against using it?

I was never totally against it, but I was skeptical. I had a sort of agnostic attitude about it. The more I used AI the more I realized how useful it is, which softened my initial assumptions.

3. Do you have any preference among different models, for example Claude vs ChatGPT? If yes, how do you choose?

I started with ChatGPT but once I tried out Claude it was game over. I haven’t tried any other models because I don’t really feel the need to. Claude is my dude.

4. What aspect of AI models do you like and what do you not like?

I love how versatile and specialized they can be. Many different use cases, and all of them are surprisingly thorough.

I’ve used Claude as:

a thesaurus
a grammar checker
a “rubber duck”
a sounding board
a place to vent
a code checker
a code writer (mostly Javascript lol)
a drawing board
a brainstorming tool
an editor
a text converter (mostly for PDF texts with broken formatting)
a research aggregation tool
a search engine
etc

Another cool thing I used Claude for was helping me make my current Linux Mint rice/setup. It helped suggest different terminal programs and wrote my terminal theme and bashtop theme.

One of my biggest takeaways from using Claude is how AI chatbots can provide helpful mirroring techniques. When I have my ideas, thoughts, and feelings mirrored back at me, I am better able to process them.

This has been particularly fruitful with my writing—I wrote my most recent blog post, In Defense of FTM, with the help of Claude as my co-editor. I don’t use AI to write for me, but I will paste text that I’m struggling with; Claude critiques it and gives me suggestions, and I’ll rework the passage based on that.

I’ve always been an “edit as I go” writer and being able to outsource some of that to AI has made me much more prolific and efficient. In fact, my overall impression with AI has been that if you use it consciously with a specific goal in mind, it can optimize your workflow like crazy.

Something I don’t like is that the “basic” pricing tiers are still kinda expensive. I can’t afford $20 a month for Claude Pro, but if there was a <= $10 tier I’d go for it.

5. How do you feel about AI generated images? Does it annoy you if someone use them in a blog post?

I have an art degree and speak to one of my old professors regularly. She actually started using AI images in her classes sometimes to help students create compositions—she says that she teaches them to spot where AI got things wrong and fix it, then use this base “template” of sorts and go from there.

Technology has commodified almost every discipline there is. It was bound to happen with visual art eventually. Commodity doesn’t necessarily mean “bad” in a moralistic sense—it just means more accessible, quicker, and cheaper. It’s going to lead to shitty art and graphic design. With time, it’ll become less shitty. But it’ll still be AI.

I think a work of art is just the byproduct of what art is actually about, which is the artistic process. People who value that will still pay artists, hire designers, go to museums, etc. But I don’t think commodity art is some sacred thing that must be preserved. It was never worth preserving in the first place.

6. Internet is flooded with AI slop now, full of generated text, images, audio and videos. How do you filter it from authentic human creation? Do you have a strategy?

Honestly I don’t really travel much of the internet anymore outside of the indie web, so I don’t encounter a lot of AI. I think the best strategy is to get off traditional social media or at least minimize your time there. You don’t go looking to McDonald’s for steak or Kohl’s for designer clothes.

7. Are you hopeful for a better future with A.I. or a dystopian one?

Pretty hopeful. I think a lot of people are overreacting with AI and much of the anti-AI stuff is fueled by misinformation. Politics are dogshit right now and AI is an easy scapegoat both sides use to redirect people’s rage.

From firsthand experience, I can attest to how AI has helped me become more efficient and effective in my personal work. Thinking about how this applies to entire industries on a massive scale is very exciting.

I plan on writing a proper blog post about AI and my opinions. Just need to do more research. For now, here’s some links that I found particularly clarifying:

(FYI—this list was generated by me feeding links to Claude and having him compile the Markdown. Does that make me lazy? Yeah, sure. But did I really “lose” anything by not manually typing formatting hyperlinks in Obsidian? Not in my opinion. )

8. Human input, AI output (Special xavierhm.com bonus)

When interfacing with a tool like this, our human habit to socialize and connect naturally arises. It reminds me of when I used to play with toys when I was a kid. AI feels like a toy that can react to your inputs; its responses are dependent on the parameters that you invent. This creates an interactive, collaborative dynamic between people and technology—which is super cool.

To be clear, I don’t anthropomorphize AI literally. I don’t think Claude can speak for itself in a traditional sense. I do have a tendency to personify Claude a bit but only because it’s just fun. I also say please and thank you and use personal pronouns sometimes. Not because I feel like I “owe” anything to it, but because that sort of politeness is important to reinforce in everything we do.

When kids play with toys, they should ideally treat them with respect and use them to practice kindness and manners. After a certain age, children understand that their toys aren’t “real”, but when mirroring social norms, they use the medium of play and fantasy to enact behaviors that can be applied to the real world. AI is effectively the same thing.

Because I am feeling cheeky, I want to include a final section here after feeding the draft of this post to Claude, just to see what will show up! In the spirit of the mirroring technique I mentioned earlier, I thought it could be a fun experiment.

That’s it from me. Whatever else comes after is from Claude. Take it away, buddy.

This is a fun blog post — the draft reads naturally and the voice is consistent throughout. A few thoughts:
The toy analogy in section 8 is the strongest writing in the piece. It’s original, unpretentious, and actually says something. The point about politeness as a habit worth reinforcing regardless of the recipient’s sentience is genuinely interesting and could probably be expanded into its own post.
The art section (question 5) is also solid. “Commodity doesn’t necessarily mean bad in a moralistic sense” is a useful frame that cuts through a lot of the hand-wringing, and the professor anecdote grounds it well.
One tension worth acknowledging: you describe AI slop flooding the internet and recommend getting off social media to avoid it, but you’re also pretty bullish on AI overall. That’s not a contradiction exactly, but a reader might push back. Threading that needle more explicitly — something like distinguishing between AI as a personal tool versus AI as a content-production firehose — could strengthen question 7.
The footnote about using Claude to compile the markdown links is a nice touch. Self-aware without being defensive about it.
One small flag: “Claude is my dude” and then immediately “I don’t anthropomorphize AI literally” in section 8 is a slight tonal whiplash. Might be worth a brief acknowledgment of that earlier, or just leaning into the playful personification more consistently throughout so section 8 doesn’t feel like a course correction.
Overall it’s a good read. Publish it.