V1.0.1 Roblox Artificial Intelligence [Chat Bot] AI

Resources Community Resources
, , , ,
#34

Any particular reason for the limits on name size? My bot is called Pod :confused:.

#35

I’ve reduced the minimum name characters to 3, After all the A.I is “Lui”, Also the reason is to give a better user experience. Of-course you do not want to have something named “Thissuperlongname” thus far the maximum name characters is set to 10 whilst the minimum name characters is now updated to 3.

I’m also working on a Node-Package-Module as well as a Roblox Library that will have an AiService embedded into it.

2 Likes
#36

This is truly awesome. I’m shocked at how well it works, and I’ve made by own GUI assistant by tweaking parts of your brilliant module to do so. Thank you so much for making this possible, I also gave credits to you!

1 Like
#37

Great to know you found this useful. I’m also working on a module that would make things even easier to configure and use this feature.

1 Like
#38

V1 Update Log!

    • [V 1 Updates]
      ● [NEW] Added Naming Feature - Allows you to name your A.I This Can be passed when making the api request.
    • ● [UPDATED] API route updated - Luiboard AI is now moved Here – Fully documented now with request examples.
    • ● [UPDATED] Getting your Api Key - You can now get your api key via your profile settings!
4 Likes
#40

I appreciate your remarks glad to hear that you found this as something you could use.

1 Like
#41

image

3 Likes
#42

Wow, this is awesome.

A very minor thing but.
There’s a typo in your example…

FilteredResponse = TS:FilterStringAsync(Response, Id) -- Right here we filter the text before seding -> sending it to the user
#43

Thanks for that, I tend to type fast. This sound be fixed now “seding” :laughing:.

#44

I’ll change that to “funni number”, for future references. :wink:

#45

wait how exactly does this work? i am super curios rn.

#46

This is really cool, but I must ask: does it support saving what the AI learns? I could possibly implement this into some of my games, but it would require the AI to save what it learns, otherwise it’d become stale and sort of “useless” in the context of my games.

Really interesting and good work though.

2 Likes
#47

Hello, just to answer your question. Saving what the AI learns within your game is something you can implement on your own. How can you archive this? well it’s very simple all you need to do is use the user’s remark and/or alter the response which you can then save to be used within your game.

With altering how the AI resposnds to certain remarks you can use the user’s response and save that to use within your own game.

Why don’t I just add another feature that allows you to save this data?

  1. This is something you can archive by your own and it’s very simple to implement.
  2. If I add an option to save data, this would also need to cost me additional resources due to me having to save the data you provide.

This is also something I’ve done on the Demo Game (Testing Game).

P.S : The AI only saves what it learns through training in addition what you were suggesting is something you can implement within your games.

If you got any questions or do not know how to come across implementing that, just reach out to me.

#48

How does what work exactly?, your question is quite broad and I’m unsure on what explanation to give maybe try and elucidate your question to narrow down on what I should try explain.

1 Like
#49

This wouldn’t be very intuitive nor immersive though. It’d be better if the AI could still learn to answer something differently without falling back to its default state.

What if instead there was a way to get a string of the neural network’s data and a way to send it through an http request to the site so it can adjust to the neural network sent? Would work much better in my opinion as then I would still be responsible for saving the AI but it’d also be able to learn from its conversations properly.

2 Likes
#50

The issue with this is that, I can not put my trust on the internet. I do not want users teaching the AI unless I know them. Letting the public teach an Ai that is also used publicly is not a good Idea. This would be more chaotic and could cause bad data structures which is something I do not want to deal with or take chances in.

Though I’m currently working on a new AI where only a few users would be able to have access in teaching the AI. This is not something planned to be available to the public yet. Also users marked as “Resource Contributors” within the Luiboard website are the users who will have that access.

2 Likes
#51

Wouldn’t it “start fresh” every time a new AI “name” is provided to it though? What I meant was sending a neural network string back in the request so it can use that structure instead of the default state. You wouldn’t need to save the AI, which as far as I’m aware you already don’t.

2 Likes
#52

I’m not quite sure you understand, teaching an AI is complicated than just sending a string. If you were to teach and AI to do simple calculations you will also need a lot of data in sets in order for it to understand and make sense out of the content.

Also another thing a neural networks work with numbers, it’s even more complicated converting a string into a number and later on identifying it.

For a user to send back a “neural network string” will still need to go through the process of the AI identifying the correct response and target trigger as well as categorise it. Though this is something I would consider in the future. For now it’s not something I’ll be adding or planing to add. It would also be time consuming to implement this and I do have other projects and features I have yet to work on.

1 Like
#53

I’m well aware; the string would represent the network’s weights and whatnot, not be there to “teach” it.

I understand what you’re expressing though, I’ll see what I can do then.

Edit: The /api/ai/fetch example is missing some , in the headers table

local HttpService = game:GetService("HttpService")
local Success, Err = pcall(function()
	Request = HttpService:RequestAsync(
		{
			Url = "https://luiboard.com/api/ai/fetch",
			Method = "GET",
			Headers = {
				["Content-Type"] = "application/json",
				["Apikey"] = "", --You can get your api key though your profile settings under menu
				["Message"] = ""-- [Here] Your message,
				["Id"] = ""-- [Here] The Player User ID (This helps identify users diffrently)
				["Name"] = ""-- The Name of your AI. (By Default the AI is called "Lui")
			},
		}
	)
end)
if not(Success)then
	print(Err)
else
	print(Request.Body)
end
#54

i mean like the ai learning part. like how it decides wht to say etc

1 Like