Predation and Pheromones in an AI Ecosystem 

I'll do my best :)

I'm loving this. If I could make some suggestions it would be to use the remains of dead bibites to fertilize new food pallets within proximity if they don't get eaten, and introduce excretion as fertilizer too. Allow carnivores and herbivores the choice to ignore each others food (or consume them based on starvation levels) Introduce a 'food poisoning' modifier that lowers speed, health and energy absorbed vs energy excreted. And maybe a poison resistance to dictate how much. The ability to stop forward motion, pivot on the spot and accelerate would be great to look at and lastly a stamina stat so the bibites can conserve or expend energy depending on their situation. P.S sorry if all this is asking to much but I love this stuff and cant help myself.

@@jamie7472 don't be sorry ! Actually, a lot of what you suggested was implemented with the last Energy update :) (especially everything about fertilizing) they can also recognise if the food is plant or meat and evolve behavior accordingly. I need to introduce more "statuses" and modifier to influence their state tho

A little late to the party, but maybe instead of changing the system to sexual reproduction, you could keep the asexual reproduction, but add the ability for others to fertilize it. (Not making it a requirement to hatch, but increasing the mutation/etc when it happens)

Interesting idea !

Thanks for the tip :) I'll try it

THIS!! I would love to see current state videos where you show a few minutes of behavior, perhaps comparing how tweaking global variables directs evolution after a few generations, videos of testing incomplete features, etc. I strongly recommend your body features update first. The others sound like a lot of work that your viewers might not be able to see. Also, please consider making the color of the Bibite a function of whatever property you are demonstrating / testing. For example, let's say you want to see what happens if you spawn "free" pellets. Make the carnivores red, herbivores green, and omnivores yellow, then let it run for a while and see if evolution favors anyone. Then tweak the numbers and run it again. Could be cool!

Problem is, that is not enough to inspire such a behaviour. A bibite has no value in it's eggs, it's something that happens automatically and does affect it's life if only as a future competitor for food or source of food or even it's own killer, but I am sure bibite's brain is not advanced enough to even come to any of those conclusions yet.

@@Traf063 I don't think so. It is kinda same as bibites own struggle for life. As it learned feeding himself and not dying to others, it can also learn to help its baby to grow and run away from others. I do not know the whole logic behind deep learning, but i think it is not so different.

@@osman4172 in order to learn that, bibites' children should be valued by them, and as of now they are not beneficial to individual bibite's behaviour. If bibites will start to learn to form some resemblance of social structures, children could become valuable as a way to increase total power of a group, but for now i don't see why they would do that.

@@Traf063 maybe they can benefit mutually when child become an adult?

it is better to be a society rather than individuals

Hmmm interesting idea

@@TheBibitesDigitalLife if you are really ambitious with the idea you could have a rotting system in place along with the pheromones. You could also eventually have plants potentially be another evolving species along with the Bibites that will grow in the rotten meat. If they get complex enough im curious if they develop farming.

A new video should be out soon, recently I had to move and the acoustic of my new room was VERY awful, so it took me some time to fix it, but I'm back on tracks ! I'm planning on keeping on and working on it for a long time, don't worry !

@@TheBibitesDigitalLife yay can't wait for the release

@@TheBibitesDigitalLife That's awesome, I can't wait to see a new video, this project is amazing

@@TheBibitesDigitalLife yay happy

😥 F

ZOOM!

Stil fast as fock boiii

Yeah, rocks or something like that would be a great addition

dmdz I think a different amount of aggression towards different colors of bibits

Loudvary McPopturn Pretty much, but with the intention of allowing the bibit equivalent of a chameleon and also the prevention of attacking a child

dmdz perhaps size could also affect speed. Small ones could move faster with less energy

xd ZwanzwaN Well, all actions would be determined by the neural network, but detecting size would definitely be an interesting concept

QeaceTheImmortal Yeah, this could be linked with the color of the bibits, which could end up with some bibits adopting a color just to prey on another bibit

With neural networks ;)

Lucian gg yes!

They will also can evolve if they can see in darkness or if they are day or night active.

or it can allow minor restructuring of neurons and memory

They should lose energy when they rest or the AI could develop the habit of permanently resting. It would also require some kind of fatigue System or else the necessity of resting wouldn't be there and a better strategy would be to stay active all the time to collect more energy.

With this he could also add a gene for noctournalism like how mice sleep at night and cats evolved to be awake at night for easy hunting

Thank you very much ! I've been working on this for pretty much 4 years (far from full-time of course) and I'm very glad ofwhere it's going :)

Agreed

@Xeno Sapien sexual reproduction, or alternatively horizontal gene transfer, is extremely important for other reasons. It allows to distribute genes among species between different lineage. As a result of that genes can spread across en entire specie instead having to outrun it

@Xeno Sapien I dont agree. Sexual reproduction has the benefit of popularizing beneficial genes, which should lead to a significant increase in fitness

That's actually a nice idea I hadn't thought about :) thanks !

I just come from a different perspective. I follow a lot of web novels and they use a similar thing for releasing chapters on Patreon and it seems to work well enough for them.

@@AkhierDragonheart that's what would make the most sense. I want people to run it for themselves and enjoy it but I still want to bring this at a serious level (I dream of giving all my time to this instead of having to work :p) so I guess it would be a good compromise :)

@@TheBibitesDigitalLife Do you have a playable version on your Patreon where people can tweak variables and test it? If not, I say do it. This would almost certainly give you more income to help free you from your work schedule.

@@blackshard641 that's the case :)

Like my channel banner says, I'm just trying to create something that's alive. But is it even possible ? What are the criteria ? Can bits be alive ?

The Bibites HMMMMM....

So true 😥

Can be avoided with a friend/foe system (and everything in between), maybe based on pheromons. If parents have one special smell, their kids should have it too, and parents would not attackt any bibit with their own smell. This smell could of course change over time if the kids grow older, and could lead them to become neutral or even enemies (maybe if one becomes a predator/carnivor, while the rest of the fam is herbivor). Chamelions then could try to imitate their enemies smell to avoid being hunted or attacked. Or predators try to gain smells from their prey or their food, maybe by following them close enough (or being close to their waste) to collect their (or other useful) pheromones, until they are covered enough in them for surprise attacks.

@@Suthriel that would work, but there would have to be a LOT of pheromons

Thanks ! Yeah I know it's been way too long :p

Camouflage would be awesome ahah

Absolutely ! I should add a genetic sameness input sense

Thanks a lot!

Thanks !

Thanks so much 😌

@@TheBibitesDigitalLife Thank you!

All great ideas ! Hatch time is already implemented (I present it in one of my earlier videos) but definitely needs some tweaking. Also, after thinking about it I'll definitely go with evolving plants ! They will do nicely with biomes and environment (varying factors like heat, humidity, etc.?) Which will affect the type of plants but also of Bibites that can live there

@@TheBibitesDigitalLife Nice to hear. Of coarse actual implementation will always be the deciding thing, over theory. And I'm not a botanist so I know very little about plant defense mechanisms to keep them from being eaten, although most I think are bio-chemical in nature. Poisons, irritants, etc, and sometimes physical things like husks or spikes, like a cactus to prevent stealing water. Come to think of it lol your bibites don't drink or use water to metabolize food do they... well if you want to go that far, the plants will probably need water, some of the animals will need water, water is a good way to have separate land masses, separate land masses would be good if you ever do include some kind of reproduction that takes locality into account... and so on. Also if your simulating a plant, many parts of a plant could be considered "energy storage" for that plant. (and overtime the plant can deplete that energy and use it for growth if a bibite does not eat it that is.) Depending on the scale, idk if you are going for more, "this is a forest" or "this is a individual tree". I've seen the "this sprite is a tree and it drops fruit" kind of thing done a lot, but there is not much to mess around with there genetically. Maybe the color of the apples change, or some fruit is edible and some fruit is not. Most of the time though they are less interesting.(imho) Maybe you could push it further and have some fruits have different "vitamins", but I think unless you have a reason how those magical vitamins got there, a reason intrinsic to the structure of the plant, then you are starting to go off the rails a bit.(but this could apply to all scales, not just big picture forest type thing) Another approach is simulating individual plants with the detail of... how would you say it... their individual limbs, stretching and growing out from where they are planted. Genetically, there is more stuff to work with there, size of leafs, where exactly on the plant the fruit grows, does the plant even have fruit,(and by fruit I mean seed + clump of energy that is easy to get, rather then digging it up like a potato) does the plant store all of its energy in one spot, or does it evenly distribute it across the structure?(bulb vs grass???) also how do plants interact with each other when they overlap? ..... and so on. Who knows, such a plant simulation might be just as interesting, or even more interesting then the bibites. :) Anyway, I'm sure your todo list is long enough as it is, so I will stop.

@@derstreber2 Not gonna read that whole thing, but if the defence of the plants are good enough, I'd imagine they could become predatory and live on decomposing meat.

@@blizzard2508-k7n I guess I assumed that as a given, but yes. It might be interesting, you could get plant formations that tend toward confusing the animals into dieing in preferred locations to then later be consumed by the plant. (I imagine something that messes with their pathing, like funnling them in, a bibit with a brain that comprehends the bigger picture of "this is a trap", would do better, etc. ) Or plants that somehow encourage animals to defecate on them. who knows?

@QeaceTheImmortal huh I should have thought of that, but no, I was think more of just regular plants in general. The phrase "pushing up roses", etc. All plants feast upon the remains of the dead. But who knows with the right pieces in place, perhaps something like a Venus flytrap or a pitcher plant or a sundew could come about without explicitly codeing a venus fly trap leaf or something.

Thanks :) I'll keep that in

Thanks a lot :) !

Would be amazing but oh boy... I have troubles keeping my computer from combusting and it's already a medium-high end machine. I don't know how it could work on a phone. Probably a lot of optimizing work

Thanks a lot ! My schedule is going to be pretty full in the next 2 weeks but I'm jumping back on it as soon as I can :)

Thanks ! And great pun

Thanks :D and welcome 👈

Thanks :D !

Thanks ! Stay tuned for more :)

Thanks :) ! I am too ahah

@@TheBibitesDigitalLife you haven't uploaded in 3 months, will there be an update soon. Also i feel like this channel could be pretty successful if you had actual regular episodes, at least once a month preferably every 2 weeks, because this stuff is pretty interesting and a lot of people like watching it. Just something to think about.

@SAQinja I'm working on a video right now (to announce the release) Yeah I too think it would be far better for my growth to upload regularly, but I can't commit to it. I have a full-time job and many other projects and activities, so I can't put too much this into this (maybe 5 to 10h per weeks max) I would LOVE to do this full-time but I'm not there yet :(

Big inspiration of course !

Xeno Sapien one thing at a time

I miss Spore. ):

Thanks a lot !

Thanks a lot for the kind comment :)

Merci!

Yeah :o it would be so cool ! But I don't know how ambitious I should be ahah

@@TheBibitesDigitalLife i think it would be best to start with letting them evolve their shape pixel by pixel. im no programmer so i dont know the true difficulty of all this but it would be nice to see variety in their shapes :)

Wow awesome ! Very nice to know you've been enjoying it :) I'll look into it, I'm not sure it's at a point where I could put it on steam yet (quality), but it sure would be a good option when I get there. I just want it to be faire for my patrons too, so I'll find a middle ground with your help

@@TheBibitesDigitalLife I understand, got to keep the patrons happy. Just throwing it out there, I think you could get a lot of traffic on steam. But take your time of course, don't want to rush a masterpiece!

Démasqué ! Merci à toi :)

Thanks !

Quite a good idea :o ! Kind of like kurzgezact with their custom birds

Yeah sometimes they are too greedy and cause their own fall

I see what you did there 👈

Why learn programming when a no-life with no friends can do it for you 🤷 ?

Yes it's awesome stuff ! I'll probably add speciation with sexual reproduction (and give them the ability to see it)

@@TheBibitesDigitalLife do you plan on making this speciation kinda hardcoded? As in, there's some kind of value that gets added to the bibites that represents their species? If so, that'd be a little boring and unrealistic. Personally, I'd prefer emergent speciation (like with my genetic similarity example), but that's just my opinion :)

@@henrycgs yeah absolutely, it's going to be based on both gene and neural similarity

I still am, dont worry! New video tonight or tomorrow I hope

Thanks a lot !

Wow, I'm not sure I even understood epistasis... I'll have to look that up further, might be an interesting avenue :o Would it be kind-of like epigenetics ?

@@TheBibitesDigitalLife Yeah, there are definitely some things in common with epigenetics! Off top of my head, you might imagine two genes, one for size and one for digestion efficiency. If you have the gene for large size, but also for low digestion efficiency, you might not grow as big as an individual with genes for large size and high digestion efficiency. So there isn't a simple one-to-one correspondence between which gene for size you have and how big you actually grow - the effect of the size gene depends on the rest of your genome. Another example that shows up all the time in the real world is gene regulatory networks, which is the most similar to epigenetics. These genes code for proteins that don't actually do anything themselves, but rather stick to specific pieces of DNA to either prevent the genes there from being read or increase how much they get read. So you can have genes whose only function is to act like switches, turning other genes on and off. And then there are higher-level genes that regulate the regulators, and so on and so forth in complicated webs. Changes in gene expression through networks like this are actually more common than changes to genes themselves, because it's often safer to tweak where and when a gene gets used than to alter the gene itself, which risks completely breaking it. It's a bit like a neural network, except for physiology rather than behavior. I have no idea how much of that would even make sense to implement given how the Bibite genetic code works, but it's just something cool that genomes can do sometimes, so I figured I'd throw it out there. So excited to have stumbled across a project like this with so much depth; it's really inspirational!

Thanks ! Working on that !

I'll do better next time ! Thank you for your support ?

Sometimes, it's fate that bring us together

@@TheBibitesDigitalLife I just came from watching your 2nd video. Awesome work so far!

You're not the first person to tell me this :o I should try it sometimes

It wouldn´t hurt at all to take a look ;) Ok, the game itself is ancient, it´s literally from the last millenium (good old games have it in their store, and some websites dedicated to its internals should still be online), but the underlying neural networks and functions are probably still state of the art, and would work in new engines too. And it would help avoid some trouble for you, because someone else already built a similar system.

Unity :)

And some coud eat it beacose they woud be imune to that poison.

Also there coud be gene that woud make them poisonous.

@@pepa6080 ohhh yeahh.. or like.. some pieces they drop are poisonous, but others are not, and they could learn to only eat the non poisonous ones..

There could be the small herbivores that might be too fast to catch or the large herbivores that are too big to eat, but energy is the limiting factor that doesn't allow creatures to be big and fast because they'd run out of energy.

Another variation to size could be growth rate, species would need more energy for a greater growth speed but in turn this would be protection from predators.

Some of the things are already the case (size and speed->higher metabolism cost, they can control growth, etc.) It all still needs tweaking tho

I'll be working on that !

@@TheBibitesDigitalLife \o/ also, to help with performance, it would be a good idea to put stuff like the pheromones, where/what the bibites are looking at, their hunger, and other similar info-providing features on toggles

good idea

Maybe the "friend factor" variable(btw float not an integer) could go negative? Maybe it could be, that if an enemy a is shared between 2 bibites, they are more likely to become friends? They could have a new output, where if a certain things happened(ie. bibites with certain color killed nearby, bibite releases certain pheromones, bibite gives food to another creature(if that gets implemented)) their friend value with that creature goes up or down. Maybe certain behaviors could be shared between bibites if the friend value was high enough? This could lead to bibites with similar likes, dislikes, friends and enemies coming together. Maybe this could be effected by / affect sexual reproduction ;)? on a whole, great idea though. :)

All great points ! Procedural sprites is definitely what I'm working on next (so many votes) Biomes and evolving plants will probably come together. Sex is also important because it allows horizontal gene transfer !

Don't worry, I didn't know how to do this at first either ! And knowing what I know now (after doing it) I wouldn't do it like this ahah. It's all about having an idea and then figuring how to do it bits by bits :) Start and keep at it and you'll end up having something amazing

Yeah, I'm probably going to work on procedural sprites first and work on doing nice code from the start so I can go a little more in depth and show it ahah. Biomes and evolving plants will probably be next

Thank you very much :) All great ideas ! Attacking others technically requires energy (it often costs health and energy has to be spent to replenish it) They don't even see eggs as of now, so It's true I should do something about that ...

@@TheBibitesDigitalLife I am working on my own simulation which is very close to yours. I would love to show it to you (I can provide a live demo) and discuss more in details about technical stuff. Can I join your discord ?

@@sylvainlasjunies4243 absolutely, DM me and I'll send you my info

@@TheBibitesDigitalLife I'm sorry I may be dumb but I can't find a way to dm you on youtube x)

@@sylvainlasjunies4243 no problem, email me at leocaussan@gmail.com

Yeah I heard about that... Still not sure how to implement that efficiently ahah I think I'll need help from a more experienced unity programmer eventually lol

@@TheBibitesDigitalLife Make a class cell/partition holding position of cell and arraylist of bibites. Each bibite also should have x and y coordinate of cell it belongs to. Partition canvas into n cells, lets say 50. Adding bibites into according cells is done by simple 4 If's. Each frame u check for each particle if it is in it's cell. If not, remove it from the cell it was in and add it to the cell that it is currentely in. So now u lose performance by all of this, but u gain tons of it when each bibite looking for closest bibite in one and neighbouring cells rather than in all of the cells.

@@likeyou3317 thanks a lot, that should help