The article describes a Google AI experiment where researchers simulated the emergence of life from non-living matter. Using a neural network, they created a virtual environment where simple agents interacted and evolved. The experiment showed that complex behaviors and even the appearance of “life-like” properties can emerge from simple rules and interactions within a complex system, potentially offering insights into the origins of life on Earth.
This keeps happening over and over and it’s tiring. Simulations leading to a randomly generated self-replicator are a) not novel and b) not sufficiently life-like. There are SO many challenging constraints on a life-like self-replicator. Each constraint is its own active area of research, and we probably aren’t even aware of most of them.
94746382926 on
How is this substantively different from Conway’s Game of life?
MaMaMaaaaa on
I mean, it is „yet to be peer reviewed“. Sounds like an undergrad thesis.
Game of Life isn’t that exciting by itself, it’s rules are simple and well known. But using it to implement a game was mind boggling.
Rarder44 on
brainfuck, even if it is an esoteric language, is not useless for this case and using c/c++ would not have had the same effects.
uses 8 characters and is easy to randomize them.
brainkfuck generally uses a different memory space for data than the code, I think they implemented a common space so that the code is overwritten at runtime by the code itself.
(I speak from direct experience as years ago I tried to implement something very similar in brainfuck)
Leave A Reply
Du musst angemeldet sein, um einen Kommentar abzugeben.
7 Kommentare
The article describes a Google AI experiment where researchers simulated the emergence of life from non-living matter. Using a neural network, they created a virtual environment where simple agents interacted and evolved. The experiment showed that complex behaviors and even the appearance of “life-like” properties can emerge from simple rules and interactions within a complex system, potentially offering insights into the origins of life on Earth.
Also, [Brainfuck](https://esolangs.org/wiki/Brainfuck) is an actual programming language.
This keeps happening over and over and it’s tiring. Simulations leading to a randomly generated self-replicator are a) not novel and b) not sufficiently life-like. There are SO many challenging constraints on a life-like self-replicator. Each constraint is its own active area of research, and we probably aren’t even aware of most of them.
How is this substantively different from Conway’s Game of life?
I mean, it is „yet to be peer reviewed“. Sounds like an undergrad thesis.
Didn’t Conway demonstrate this a long time ago with, you know, the [Game of Life](https://en.m.wikipedia.org/wiki/Conway%27s_Game_of_Life)?
I’m still more impressed with the project to create [Tetris with the Game of Life](https://codegolf.stackexchange.com/questions/11880/build-a-working-game-of-tetris-in-conways-game-of-life).
Game of Life isn’t that exciting by itself, it’s rules are simple and well known. But using it to implement a game was mind boggling.
brainfuck, even if it is an esoteric language, is not useless for this case and using c/c++ would not have had the same effects.
uses 8 characters and is easy to randomize them.
brainkfuck generally uses a different memory space for data than the code, I think they implemented a common space so that the code is overwritten at runtime by the code itself.
(I speak from direct experience as years ago I tried to implement something very similar in brainfuck)