A few weeks ago I filled out a Canadian government survey on generative AI in the arts. (The survery is still up as of this blog post, in case any Canadian readers want to fill it out.)
The survey was hilariously biased towards AI (questions are all like “What help does your organization need in order to adopt more AI?”), and by the time I got to the free text entry questions at the end I was pretty pissed off.
But one question, in amongst a list of questions about how your organization currently uses AI, was “Does your organization have an AI policy?” I initially clicked “no” to this along with all the other questions, but later on went back and changed it to “yes”, because I do have a clear gen-AI policy, although I hadn't written it down. My policy is that I don't use it.
So for the record:
Space Goblin Games Gen-AI Policy
I will never use generative AI in any capacity.
I do not consent to the use of my work in any project that uses generative AI.
I always try to say “Generative AI” rather than just “AI” when talking about ChatGPT and similar things. “Artificial Intelligence” is a broad, decades-old academic field, and also (as in Seedship and some of my other games) a science fiction conceit about “what if a computer could think like a person?” Generative AI is a sophisticated autocomplete that uses ecologically devastating amounts of power and water to create text and images that are statistically similar to text and images in its training data, which was scraped from the internet and from pirated book archives with no regard for copyright or consent. Calling it “AI” is a way of confusing people into thinking it's like a science fiction AI, which it is not. It's very often wrong; it's been shown to cause cognitive decline in regular users; and it's being used to put people out of work by bosses who think it can do workers' jobs.
When used in creative work, gen-AI's outputs are also soulless and artistically bankrupt. Gen-AI is good at text that seems OK when you skim-read it but has no deeper meaning, and images that look OK at first glance but fall down when you look at them closely. If someone's impressed by gen-AI output then I've got to assume they've never really thought about a piece of text or looked closely at an artwork.
I've seen some people online do things like get ChatGPT to play Seedship, or having it output stories or images based on their Seedship games. And, like, I can't stop people from doing that kind of thing, but I want it on record that I would prefer they didn't.