Use Randomizers to Mimic Machine Learning

In Gimkit creative, people have created randomizers. And yes, I know that they’ve been around for a while, so why am I bringing this up? Well, I thought that you could theoretically code an AI using randomizers. And yes, this isn’t really an AI, but it’s probably the closest thing we have to an AI for a while. So, what you would do, is you would have one randomizer that could choose between 4-6 actions it could perform, so it’s a bit simple. When one is selected, it broadcasts to another randomizer that chooses another action. In short, you choose actions for something, like, a sentry to perform. Maybe it moves. Maybe it just disappears. Maybe it says something. My point is, if you have a long chain of randomizers for the game to basically choose actions for itself to perform, you’d get an AI. It really may not seem like it, but it’s probably the best we can do for awhile. Also, I made this a guide because it seems like it fits well here. I also had more stuff to say on this subject, but I want your guys’ thoughts on this, and I forgot some of the things I was going to say. Also, credit to this guide: https://forum.creative.gimkit.com/t/how-to-make-randomizer/711
(By the way, in the guide, he mentions you need a repeater. You actually need a trigger. Just thought I should let people know.)

Edit - This really isn’t AI, so I changed the title. I based it off another comment that said it’s basically machine learning.

14 Likes

Hmmm…. put this under concept

3 Likes

@Shdwy, what do you think of this?

5 Likes

I don’t think it classifies under AI but it is a good representation of it with randomizers helping simulate AI.

5 Likes

That’s what I thought, it would basically just be a Walmart knock-off of an AI

7 Likes

Nice concept guide!

7 Likes

Makes sense!

Machine learning seems pretty feasible, even in the context of GKC. You wouldn’t be able to do the traditional reinforcement learning neural network style here, but you could absolutely have a decision matrix that updates based on success rate or rewards.

That’s actually really cool! Can I put this in my technical gimkit post?

5 Likes

Yeah, that’s fine! Also, I think this is my last reply I’m able to post today, so if it is, email me for further discussions.

3 Likes

We are quite literally getting into the matrix of gimkit technology.

4 Likes

Oh, well, gee, thank you!

1 Like

Not really an AI but more of a computer player controlled by random actions.

2 Likes

no because even though deep learning could be a gimkit concept, this is just using a randomizer which is an ineffective means to the problem and can’t revolutionize (it smth we alr know put in a different context), this is essentially the bogosort to ai

3 Likes

Yeah, some people have said similar stuff like that. I know it’s really not an AI, but it simulates it, and it’s the closest thing we have right now to it.

1 Like

ai is a simulation, so you’re simulating a simulation, and even so, ai is possible, just complex (as ai is irl) so it isn’t really the closest thing /nm

4 Likes

it is a nice simple approach to the problem though so ill give you that

3 Likes

/nm? as in never mind or is that a typo

1 Like

tone indicators, there was someone who explained it here before i but i forgot where it was

3 Likes

Well if you have thousands of options for the game to choose, I guess you would have a prototype of an AI? Or something even lesser than that.

1 Like

I’ll try and start working on this soon, feel free to give me any idea about what I should code it to do

1 Like

Yeah making a simple neural net is 100% possible

1 Like