The crowning achievement of my PhD was having some of the images I took featured in one of Ze Frank's videos 10 years later. Who cares about Science and Nature papers!
Ubelievably thought provoking. If it can solve shortest paths, and it can learn, I wonder if they can solve other graph problems, then I'd wonder if there is some fusion of conway's game of life and transformer models that might produce something like a logical slime mold, or a way of integrating training sets for ML models. If life were the expression of a kind of architecture exemplified by the minimum features of these slime molds, and then this may be synthesizable programatically like we do neurons, we'd be into some interesting territory.
>Fuligo septica is a species of slime mold, and a member of the class Myxomycetes. It is commonly known as the scrambled egg slime, or flowers of tan[2] because of its peculiar yellowish, bile-colored appearance. Also known as the dog vomit slime mold, it is common with a worldwide distribution, and it is often found on bark mulch in urban areas after heavy rain or excessive watering. Their spores are produced on or in aerial sporangia and are spread by wind.
Sebastian Lague made an excellent video "Ant and Slime Simulations" which explores how path-finding algorithms can use trails and gradient-based signaling to solve spatial problems: https://www.youtube.com/watch?v=X-iSQQgOd1A
This spring I have been collecting pictures of the various fungi and slime molds in my yard. One of the stunnning things is how quickly they can change from slimy and wet to dried out and desiccated. In two days one went from wet bubbles (10:38 in the video) to dust on little stems (10:50). The time lapse pictures make me want to set up a camera and record the next one I find.
Here are some very cool images of Myxomycetes. The metallic looking species are very cool, I'd like to see one sometime.
https://www.myxotropic.org/gallery/
