Current projects

I'm generally working in the area of trying to merge open-endedness, artificial creativity, and complex systems.

I may update this post occasionally as I learn things, and will probably shift projects occasionally if things don't pan out.

I'll try to have a policy of openly sharing my ideas because I don't care if I'm scooped. I just want cool stuff to be built, I don't care who does it (though don't worry, when I collaborate with others I don't use this policy because I understand others feel differently).

My goal is to figure out if there's a way to make systems generate open-ended content that seems "human culture-like" in it's form. While I like Stanley's approach and will use many of the insights of POET and it's follow ups, just body locomotion seems like it'll cap out in a way that's distinct from culture. Maybe there's a way to transfer via some reduction, but I'd like to see if it's possible to push in more text-like and system-like directions and somehow get around the issue of interpretability (the fact you can easily see what the creatures are doing is one of the main reasons he currently focuses on locomotion, wheras settings with language, programs, and theorems get really hard to understand what's going on, which makes it hard to diagnose).

The two directions I'm currently pushing in are:

- An economy/industry sim, where individual products are theorems (using metamath). The goal is to try and have some open-ended economy type behavior that matches many of the empirical laws of innovation. The main goal is to have drift: I don't actually really care which particular theorems it focuses on, I just want the economy sim to value different things and continually produce and invent new things and never really stagnate. 

- Discovering interesting systems: The goal is have an agent working in Conways Game of Life that tries to discover interesting things. To do this, I'm having one agent try and produce things where it's difficult to predict what it'll do, and another agent try and predict what'll happen. This is tricky to balance right because if the agent is too expressive it can just learn the rules, so it has to be forced to compress it in some way, but I think there's ways to get around this. If I can get agents discovering interesting things, that acts as a good measure for an interesting system. Then I can generate lots of systems (mostly cellular automata at first), and give them to an agent, and see how it reacts to those systems. 

Part of this also involves prototyping systems to get a better feel for the design space, because I'm still a bit of a game design noob. I've also been reading Rules of Play and have found it pretty informative.

Comments

Popular posts from this blog

My prediction of AGI in 2021-2023

Feedback Loops in Opinion Formation

Inductive Linguistic Biases