What happens after 1000 hours of Evolution? Recreating the largest evolution experiment ever

In this video, we dive deep into the fascinating world of The Bibites by recreating one of the most iconic real-world evolution experiments. Inspired by the famous Long-Term Evolution Experiment with bacteria, our simulation runs through thousands of generations of digital lifeforms. These Bibites start with simple behaviors and genetic traits, but as they live, eat, reproduce, and mutate, we witness the power of evolution in action.

Just like in the original experiment, where bacteria evolved over 75,000 generations to adapt to their environment, our Bibites face their own challenges. Through natural selection, they gradually evolve new traits, form distinct species, and adapt to their surroundings in surprising ways. By the end, we'll have not only new species but an entirely different digital ecosystem, demonstrating how even small genetic changes can lead to significant evolutionary leaps.

=================================

The Bibites is an Artificial Life (Living AI!) simulation where I recreate some biological processes and let the lifeforms live, eat, reproduce, and mutate, leading to active evolution.
They can evolve their body through a genetic algorithm and their behavior through a custom neural network algorithm.

=================================



=================================
Credits:

#evolution #AI #devlog #simulation #ecosystem

[zenlis]

in order to have bibites that interact with the environment like beavers or primates, you have to first establish plant evolution. plants with armor, plants that kill them, plants that give more or less food, even invasive plants would definitely help achieve that goal. overall, i really enjoy watching this project grow and grow! keep it up! :)

also eyyy i made it in the vid as comment lol!

[gem]

The best evolution simulators are the ones that are built line by line. It's clear you have been METICULOUS with this game, and I hope you're proud of yourself.

[Juaels]

36:10 It's so funny, you got "only" 1800 generation (42 times less than the lab), but it took you "only" 10 months, which is exactly 42 times less time than their 35 years. Both your artificial life and real bacteria evolve at the same rate, which is crazy! Great job 😊

[alessandrosilvafilho]

One thing I think this simulator needs is more niche possibilities, such as different plant parts like fruit, wood, leaves, and roots, or maybe things like rotting meat

[aproductions]

The behavior of these bibites blew my mind! Evolving red skin to disrupt pheromone production, using pheromones to mob foreign species, taking turns eating, running away before giving birth, being less antisocial when starving... That's SO COOL!! It's come so far since I last downloaded it, and the new species analysis tools will make tracking these longterm sims so much easier.

[Skarix]

Naming the Bibite species after patrons is such a cool and funny idea. Wacky sciency names is so on-brand for this project

[sebastianfries]

You should reach out to the developer of “the sapling” I think that collab would be legendary.

[alfieho-brcw]

we are starved for content, master
give us your artificial life

[krispyso]

this is genuinely such a cool project. im so happy to see this silly project evolve and become more complex, even if it takes 10 more months for it to be reflected in the channel lol

[darkjill]

Oh shit he's still around I figured he found happiness and went back to nature.

[anthonycannet]

I have an idea for keeping the simulation from ramping beyond your computer's capabilities. If you tie the rate of generating plant material to the current simulation biomass you can set a hard limit to the amount of biomass and by extension the number of entities in the simulation. For example if the biomass is limited to 10, 000 then as the simulation approaches a high number of bibites, the amount of plant food sources decreases. On the opposite side, meat food sources should deteriorate over time to return the available biomass to creating more plants as the bibites die. In theory, this would make a more strict limit on the population than competing over nearby resources but would more closely represent how it works in real life, the amount of matter on earth doesn't increase. The available organic matter might go up from microscopic life consuming inorganic material, but that just shifts the "biomass limit" to the available inorganic material for the bottom of the food chain.
Basically, as the population starts to reach the limit of computing power, the food generation decreases to prevent the simulation from expanding exponentially.

[tellder]

12:40 it's so funny that the bibit gave birth, and the newborn immediately attacked it's parent

[arnautarnautsen]

At 34:25 you hit the main problem of all life simulators: diversity in actual biological systems is due to the fact that new species *become the environment*, sometimes literally, as in endoparasitism, and sometimes as a mass effect, as in trees or corals. Failing to implement this results in something that, more than an evolutionary simulation, is an evolutionary algorithm, and those converge to a single solution when given a single set of boundary parameters. This particular simulation promotes diversity by artificially introducing different environments, which certainly is a factor in biological diversity, but tends to simply produce a single convergence of the algorithm for each environment.

[arhanrahi]

Mom: my son will be a biologist!
Dad: No, he will be a programmer!
The son:









Jokes aside, this is a really cool project. Been watching since start how you continue man!

[callmebeep]

You should add a function to that game where pheremones drift in space in a direction pattern controlled by the settings of the game, and also that the plant and meat pellets each have their own value of "pheromone radiation". I understand this may have a performance factor, but this allows scavenger/lower-density bibites to have another way to track down food and evolve a more modified niche.

[fabricreative]

It's so cool how these creatures find unique adaptations to solve problems! The cladogram really came in handy in this simulation!

[Mastervitro]

Imagine running their evolution with super computers the same way we do with A.I.

[燧發槍]

One idea it might be good to add is individual memory, memories are passed down in evolution but if a bibite looks away from food for one tenth of a second it has no memory of it at all, a basic short term memory might be a neat ability to evolve, this could also help with them wandering off because they could remember in what general direction the food is

[vonnegutfrey]

The Sapling Evolutiongame simulator recently added sexual reproduction, and it became a real game changer there. I'm curious to see what it would do in the more detailed Bibits simulation.

[sips]

hello! concerning your predator-prey bias problem (if it hasn't already been fixed), I theorize that having fitness landscapes will help. something with the current diet system, is that I think it's A to B, 0 to 1, a slider like most properties of bibites I think. The real world takes several steps to change things, like a diet. Teeth, chewing method, acidity of stomach acid, amount of stomachs, etc. Do your own research as I have not delved into this too much. May bibites survive - and you thrive!!!