Max Cahill

@dougmerritt@mathstodon.xyz @demofox@mastodon.gamedev.place love genetic algorithms and beam searches and the like. one of my first ludum dare games was a genetic algorithm where the fitness of a specimen was taken as its average distance and minimum to the player over its lifetime and its genes were a stream of inputs, and they'd gradually learn how to get close to wherever the player was building. was exhilarating to see them learn in real time. fwiw i am extremely anti "AI" - the current trends are a huge shame though because theres a lot of very interesting weird cool not-awful ML/NN stuff to explore, but LLMs have pooped in that particular pool.

@dougmerritt@mathstodon.xyz @demofox@mastodon.gamedev.place maybe! I kinda think society is chasing the dragon there and the wheels will fall off if everyone actually has to pay for it rather than just speculating. We'll see!

@demofox@mastodon.gamedev.place

It seems like mixing multiple parameters to optimize should be some non linear thing.

I mean, they are, thats why you have to step it and re-evaluate and do so a million times, and why weights can blow up, haha

But yeah it is kinda wild magic. Just attribute the change all the way back through the entire thing and then smoosh the numbers a little better

@beep@follow.ethanmarcotte.com @sboots@mastodon.sboots.ca excellent write up and yeah wow that's a lotta links haha

@bohwaz thank you for tipping me off to this fantastic feature 👌 another reason to love sqlite