Feyntan (@Feyntan) header
Feyntan (@Feyntan) profile picture
Joined April 2013
Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)Photo by Feyntan (@Feyntan)
Feyntan retweeted
テツノブジン18連
2
563
6
3,543
35,412
Feyntan retweeted
着物と琉装のサーナイトを置いていく
7
958
6
7,059
54,930
The world needs more people like Grant Sanderson
226
[A HOUSEKI INTROSPECTION] Small Phos and Aculeatus fanart #宝石の国 #housekinokuni_fanart
2
184
1
887
6,995
Feyntan retweeted
2
164
2
1,518
11,724
This may as well be a throwaway account. I’m generally disconnected from my past, but this is the only place where I had an audience. But I think what I really need to do is to just try engaging in academic communities. Hell, maybe I should go back to school.
1
3
539
People here knew me for making games, but I hardly even play games anymore, and now I only make games when I get paid to.
340
I don’t think open-ended evolution is taken seriously as an approach to intelligence, and probably for good reason. I don’t know if it’s even remotely feasible, but I would find it more satisfying than intelligence made by distilling human data.
1
2
318
I also just think evolution is really interesting, and I want to demonstrate intelligence that emerges more naturally, like our own.
1
150
Feyntan retweeted
ラメ加筆✨ そのうちBOOTHに追加します💎
1
138
1
1,181
8,002
These days, I can have Claude write and run experiments, then report back with results. I can iterate quickly on all my stupid ideas. And it can be doing stuff in CUDA on my PC or Metal on my Mac. It's nuts, but it does make me kind of sad to not be doing much of the work myself.
1
332
Last year, I got kind of obsessed with the idea of a keyboard based on hand tracking. It didn't work as well as I'd hoped, but I still think it was a neat idea. It's called ProprioType, and it's on the App Store, for the few people with a Vision Pro who might be interested.
2
214
A couple years ago, I was messing around with Niagara in UE5 to basically experiment with some weird algorithm ideas I wanted to try. That was basically my first time doing compute shaders, and UE5 was what I was already familiar with, so I tried to make it work.
1
229
I also tried to get the same architecture to control a little softbody critter, to see if they could evolve to follow a dot on the floor. Each leg had three muscles, and each foot could sense distance to the dot. It sort of worked lol. youtube.com/watch?v=vdxgcKgq…
1
57
Ultimately it was kind of stupid, and Niagara is really not ideal for this kind of thing anyway, so I gave up on it, but I still enjoyed experimenting with it. Compute shaders are fun.
31
My main interest these days is stuff like ALife and evolutionary robotics. I'm particularly interested in continual local learning, like Hebbian plasticity and NCA and stuff like that. I'm not really in academia, but I like learning and trying to make my own simulations.
1
229
I basically left social media a few years ago, around when AI had started to really enter the public eye. It felt like everyone I knew turned violently hostile toward an entire field of study that I had taken a keen interest in, so I kinda just peaced out.
2
13
1,124
It’s been nice just keeping to myself and focusing on things that interest me, but I think I might have gotten too comfortable with not talking to people, so I’d like to try meeting people who I can actually talk to about interesting things.
2
3
269