Sakana AI’s God Simulator Is Brilliant

Hinzugefügt: 2026-05-08

[00:00:00]I love this work. This comes from the Sakana  AI lab in Tokyo, and it feels like mad science in the best possible way. I had an almost illegal  amount of fun with it, and you can try it too right now. Assuming we don’t crash this website,  which we sometimes do in a Scholarly Stampede.

[00:00:20]Now this is wild - we are going to  play God in a tiny digital universe.

[00:00:27]Look, we have 5 AI species in a petri dish  fighting for territory and…nothing. Why is that?

[00:00:36]Well, this world is brutal We made the  threshold for survival sit too high.

[00:00:42]The environment crushes them. And we can  run the simulation for as long as we want, no luck - none of these  species can get a foothold.

[00:00:50]It’s a bit like the app store.  Thousands of new apps launch, almost none get traction, and most disappear  without a trace. It is simply too difficult to stay alive. AI wrapper companies?  Killer new social media apps? Same deal.

[00:01:07]Okay, enough of this. Let’s be a good god.  Let’s make life a bit easier. The market is be suddenly flooded with money for AI startups.  Have a crappy idea? No demo? No business model?

[00:01:22]Great, here’s a billion dollars brother.  Let’s gooo! So what happens? Look! Whoa!

[00:01:34]Empires start to grow and  grow…oh! But too quickly!

[00:01:41]These empires grow out of nothing  quickly, and then crumble just as quickly.

[00:01:46]Why? Because the bar for survival is too low.

[00:01:51]So we tighten the screws again. Easy money  dries up. We get a tougher economy, and…look.

[00:02:00]All these companies that got  addicted to the easy money, suddenly disintegrate.  Gone. And then…what happens?

[00:02:08]Look! New ones appear that are more  adapted to this environment. And as we start playing with some of these  variables, we change the environment, and can almost even create one ugly  monopoly, or an unstable ecosystem.

[00:02:25]I love this, this is so cool! So how does  this work? Well, there are some basic rules.

[00:02:33]The paper calls these species neural  cellular automata. This is a living, trainable pixel world inside a 2D grid. It gets  pretty involved, but here is the gist of it: each organism competes for these pixels  they can grow from nearby territory, and must stay alive by local support. But  here is the key: they are constantly learning.

[00:02:57]Oh, and they can also fight! They  specify which direction to attack in, and which direction to defend. Kind of like having  a sword and a shield pointed to a direction.

[00:03:09]In the actual work, this happens in a higher  dimension, you see a simplification of that here.

[00:03:15]When species meet at a border, the system  calculates which cell defeats which other one.

[00:03:20]What I absolutely love about this work is  that whether we create a world of total chaos, or a stable and healthy  ecosystem. It only depends on us.

[00:03:32]Change the environment, and you change the  winners. It decides who rises, who falls, and who never gets a chance. Nature  works like this. Markets work like this.

[00:03:44]Everything works like this.

[00:03:45]This really shows how important it is  that we have good policy in a country too.

[00:03:51]Most of the time, you let competition  do its work and find the winners.

[00:03:55]But sometimes, one tiny push  makes the whole system healthier.

[00:03:59]But it gets even more beautiful. We talked  about competition, but what about collaboration?

[00:04:06]Is that also possible? I thought of  course not, because each organism has one single objective: grow. This creates extreme competition, not collaboration, right? Well, I was completely wrong.

[00:04:22]Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér.

[00:04:26]Step number one. They call it permissive  mixing, I call it growth. We create an environment that is very forgiving. Digital  species run wild and spread out everywhere.

[00:04:39]Then they bump into each other constantly.  We get a big soup. No firm borders.

[00:04:45]Step number two. Crystallization. We raise  the threshold a bit. The rules get stricter.

[00:04:53]To survive this harsh world, species must group  up into dense, solid shapes to stay alive.

[00:05:00]Competition draws the borders. Everyone finds  out who can live next to whom. Borders harden.

[00:05:07]Then, step number three. Relaxation. We ease  into a more forgiving market. Now what? Wow, this is amazing! We get stripes, and little  checkerboard patterns between these empires.

[00:05:23]Why? What happened?

[00:05:25]Well, in the previous stage, weak border cells  died immediately. It creates borders like when two colors of paint drying while there is a  masking tape between them. Colors separate. Now, hold on to your papers Fellow Scholars, because  in stage three, we remove the masking tape.

[00:05:51]What happens? The edges break open. They flow  into each other, creating these little streaks.

[00:05:59]Yup. These empires are forced to coexist  because the game is unable to kill weak border cells so easily. Both sides  can keep tiny pieces of land at the edge, instead of one side fully erasing the other.

[00:06:13]Beautiful. Simple. Brilliant. But this  is also an incredible life lesson.

[00:06:19]If you are always too loose,  your life becomes a soup.

[00:06:23]If you are always too strict,  your life becomes a prison.

[00:06:28]I am guilty of this too. This is not the way.

[00:06:31]Instead, start out loose, give yourself room to  find your way. Then, build discipline. Harden up, find your shape. Establish boundaries. And finally, to not stay frozen forever, loosen up a bit. Grow. Adapt. Let new things in.

[00:06:52]Beautiful life advice in a beautiful paper.  Try it out for free in the video description, it is amazing. What a time to be alive! If you enjoyed this, subscribe and hit the bell so you get more papers. And  more papers is always better.