My AI manifesto

This is my personal manifesto for how I will and will not use generative AI. I believe it can be an extremely useful tool in certain situations, but can also have many negative repurcussions if used incorrectly.

1: Use to expand my abilities, not as a crutch.

A: Teach me new things

As a coding assistant, AI can quickly show me the correct syntax. It can suggest implementations, and teach me how to implement new functions. These are extremely useful, and help me to be exponentionally more productive as a hobby developer.

I must take care to use this as a tool to help me learn and grow. Instead of receiving the AI’s solution and calling it done, I need to understand what it has suggested and why, so that next time I will be better able to produce that answer on my own.

B: Do things I can’t do

If there is a trivial thing that I actually can’t do, and it is a progress blocker, I will use AI to fill in the gaps in my skill. Drawing a logo image for a silly pet project, creating a colorful, abstract background image, etc. It doesn’t make sense for me to learn how to do that for myself, so I feel comfortable delegating that to AI.

C: Don’t do things I can do

I do not use AI to write for me. I am able to write for myself, to express my thoughts through words in order to share my ideas with others. I believe that skills deteriorate without practice, and I am afraid of ever losing that ability. I will never delegate my writing, my thinking, and my shaping of ideas.

2: Treat as a tool for creators, not as a creator unto itself.

I work in a very creative, collaborative environment so this is very relevant professionally as well. It’s possible for me have an AI generate ideas, texts, and images, and by doing make me seem self-sufficient without collaborators. I will not do that.

A: Help those who create

There are many ways for creative professionals to use AI as a tool within their process. If an artist uses AI somehow to help in the creation of their own art, that feels fine. As above, my hope is that they will use it as a way to expand their capabilities, and not as a crutch.

B: Don’t try to be a creator

It feels wrong for a non-artist to generate images using AI and treat it as creative output. If I need visual artwork as part of a product, it will be created by a human creator. AI creations will not replace human creations.

AI can be used as a tool through the creative process, but will not be directly creating any part of the finished product. That is the responsibility of the creators, the people, and is too important to be delegated.

C: Help me to collaborate

While I will always value my ability to communicate through words, sometimes an image can be far more efficient. Finding common ground through concept art, such as by generating many sample images to collectively define what is needed, can be a fast way to reach a shared understanding.

The ease of producing many different formats for presenting an idea allows for experimentation in communication, which has possibilities that aren’t even fully explored yet.

Corollaries

1B + 2B: Do what I can’t do, while not replacing real creators

I will use AI to do things that I cannot do, but I will not treat the AI as a creator. Anything I create that directly uses something generated by AI will be presented as a proof-of-concept. I will not treat it as a finished product.

Anything that I present as a finished product will be ultimately created by people who can take responsibility as creators. Delegating that role to an AI is irresponsible.

I may have just flipped a setting that will halve our utility bill. Equal parts happy that I discovered this, and annoyed that I had been pissing away money for so long.

The Photo Composition Challenge

I am a mediocre photographer at best, but of course I want to improve. When I saw that @hiro was putting together 5-week challenge to experiment with various compositional techniques and share our photos, I had to join.

I don’t have fancy gear, or a lot of time to go hunting for photos, so almost all of my shots were scenes from my daily life, taken with an iPhone. I realized pretty quickly though, that being deliberate about composition gave a new perspective on even familiar scenes, and I was surprised at compelling some of the photos turned out.

Continue reading →

Sage words from the 4 year old:

You need to be the first to tell mommy that she’s pretty. Say it first thing in the morning, as soon as she wakes up. That’s the kind of adult _I_ want to be

So what’s the best way to feed my car’s user manual to an LLM and have it answer all my questions? Or every appliance in my house, for that matter.

Python script failing via cron

The issue

I had setup a cron task to run my Squirrel Archiver script once per day, but for some reason the task was not executing. Nothing, including any errors, was showing up in any logs. At first I thought there was something wrong with my crontab, but other scripts set up to run in the same way were running without issue.

Copying the commands in crontab to the terminal and running it manually worked. It was only when cron was triggering the run that it failed.

Just to be sure, I changed my script to just print('hey'), and this worked! So there must have been something wrong inside the script, and I started the tried and true debugging process of deleting chunks of code until something works.

Side note: Since the script had to run via cron, while I was debugging it was set to run once per minute on the dot. I had to rush to make changes in time, but I also had to wait to see results, which made for an interesting experience.

The culprit

I had been declaring openai_client = OpenAI() as a global variable, outside of any function. This worked fine when running the script manually, but when cron triggered it, it failed silently.

Why?
I have no idea.

Is initializing objects outside of functions commonly known to be a bad practice?
Probably.

I wish I knew more about the reasons, but moving the declaration inside a function fixed the issue.

So, maybe to many it is obvious that “yes of course you shouldn’t do that”, but I’m recording it here in case anybody else out there is like me.

What I'm doing /now

Personal Life

Nothing much has changed for our family these past few years, and that is by design. Work is stable, life is stable, everybody is happy. I am very thankful for that.

Work

Since 2022 I have been leading a department that is tasked with making our studio better at creating new games. This involves encouraging all staff to submit game ideas at any time, working with people to strengthen game ideas and create attractive pitches, coaching teams through early development phases, maintaining a corporate strategy for the release of new games, and more. It allows me to work directly with a lot of people throughout the company, and has been tremendously fun.

Hobbies

In 2022 I joined a Haiku study group. We decide on a few prompts every month and submit our poems in advance of a meet-up, where we comment on each other’s pieces and have our teacher critique them. Haiku is fascinating because its extreme minimalism (only 17 characters allowed) makes it feel like putting together a puzzle. It is also a great way to learn about nature and Japanese culture.

Last year I set up GoToSocial as a solo instance to join the Fediverse, and that has been a lot of fun. It’s nice to be social on the internet again. I think this has helped motivate me to update this site more, and I’m having a lot of fun making little updates to the site as well.

AI tools have been super useful for programming as a hobby. It has made it so much easier to finish and ship my little projects, and I have already developed a slew of little utilities that I use often. I’m still writing words by hand though, because I think being able to organize my thoughts into communicative sentences is such an important skill.

I also started keeping a physical journal at the beginning of this year, and one month is the longest I have ever kept up the habit so I am hopeful that it will continue. I had always admired people who keep physical journals; I thought it was a wonderful habit. I never could get into it though, and had resigned myself to thinking that I just wasn’t that kind of person. But my wife convinced me that if that is the kind of person I want to be, then it’s worth putting in the effort to make the change.
So far, I am really enjoying it. I’m using the most simple interpretation of Bullet Journal. It’s turning to be a super helpful framework to guide the kinds of information I should keep, and the format for recording it.

Katamari Damacy postmortem

By Game Developer Magazine:

We recently received a request to publish the Katamari Damacy postmortem from the December 2004 issue of Game Developer. It has been posted here in full for the first time in 2024 to celebrate the game’s 20 year anniversary. Please enjoy.

This article was written by the game’s director, Keita Takahashi, and has been translated into English from the original Japanese. I wish I could find the original text, but this is a great read regardless.

It’s easy to take a successful game and say “Here are its unique features and the reasons why we think they led to success.” But this portmortem also includes regrets from the game’s development, which gives a really rare, close look into what the team’s ambitions were and how they thought about the game’s design.