Um, what's up with the blurred out sections on the screen? I get there's a link that's no longer good at the end, but it covers up speakers names, Twitter handles, slide images etc for a lot of this video
In the process of designing a game following some of these principals. I’m really curious to see if the bigger limitation is the designers ability to anticipate and effectively codify all the various motivations and influencing variables, or the computational power to actively simulate all these variables
@@AZTECMAN she is not "a person", she is the other realtor. so I think the blur it's pointless if it's for that. I think it's for something else. 14:30
Seems like the narrative interactions would be good to crowd source, just send out lots of random verbs and objects and ask people to give suggestions of what they lead to, do that a billion times and you'd have good data for an AI...
Not really. I've just taken a machine learning class, and done some external reading. I don't know if there's anything in particular you had in mind, but I'm aware of recurrent neural networks.
Controlled generativity is in a much cruder state in neural networks. They are excellent and (mostly) unbeatable for classification and regression, with the proviso that you have a ton of data to train it. But it's not, and likely will not, be the be all end all of AI. Even the researchers who developed the techniques that overcame the problems training large neural networks admit that they have no idea of *why* and *how* the resulting trained models do what they do. Deep neural networks in summary. To deal with noisy data from the real world: great. To generate behaviour with any remote semblance of control over what is generated: years away, if at all possible.
well thats looks very likely HTN planners, already they are very abstrac, just define task, and will get compund task. Maybe i will try this some day but actually i thoght its better a auto-narrative gameplay. Dark souls dont need tell nothing you can always deduct what happened.
Um, what's up with the blurred out sections on the screen? I get there's a link that's no longer good at the end, but it covers up speakers names, Twitter handles, slide images etc for a lot of this video
I love the cross-section of engineering and art.
Dwarf Fortress: The most namedropped game regarding procedural generation.
and deservedly so :D
I think you might be misunderstanding what this video is talking about.
@@JTX30000 At 2:25 Dwarf Fortress is name dropped and shown. How did he misunderstand again?
The dove flew down to the stream.
The ant insulted the dove.
The dove ate the ant.
_A kindness is never wasted._
literally how a primitive story generator would work.
@@javedakhtar1435 the dove got a feast, not wasted
@@NeoShameMan bro i don't even remember when i put this comment dpwn lol
@@javedakhtar1435 that's the fun of this
In the process of designing a game following some of these principals.
I’m really curious to see if the bigger limitation is the designers ability to anticipate and effectively codify all the various motivations and influencing variables, or the computational power to actively simulate all these variables
How's the game designing going? Did you learn something interesting?
Caves of Qud is another really good proceduraly genrated game
Why is it that halfway through this video some parts are blurred out?
I assume it's the link being blurred out because it's not safe
MadsterV that makes scenes.
If you look a 20:40, there is a person's face behind the blur.
@@AZTECMAN Nice!
@@AZTECMAN she is not "a person", she is the other realtor. so I think the blur it's pointless if it's for that. I think it's for something else. 14:30
Seems like the narrative interactions would be good to crowd source, just send out lots of random verbs and objects and ask people to give suggestions of what they lead to, do that a billion times and you'd have good data for an AI...
Look up GPT-3. its an AI built just like that, but instead they just used a massive amount of text from the internet.
Its very impressive.
Happy to see the thumbnail for this is from the old Sinclair Spectrum “The Hobbit” game!
(at Darmok and Jalad, Tanagra)
shaka(A and B) = (shaka A) or (shaka B)
I really want to see more games using this.
I'm working on it...
I am working on one in cpp console it won't be big but it will deeply intricate and complex. sort of like a life simulator.
@@feelsunbreeze Anywhere I can follow the development?
Why is the presentation censored?
I'm really curious where these eather novel deep learning language models will take this!
so, this is basically logical programming like in prolog, no?
Cool!
My video had censor bars on it .. ??
WHAT THE HECK IS THAT BLUR
a bad url
no kind effort is wasted... great philosophy
Why are parts of the video redacted?
It had a naughty word
the blurring is super weird.
7:52 Is this logic programming?
Reminds me of Prolog as well.
It's nice to see these old ideas implemented in games.
Yeah I've been thinking about using core.logic in clojurescript for games. Interesting though to use it for the narrative in the game.
On GitHub, there was some preliminary work by Chris and some Scheme/Kanren people about doing this in a system similar to core.logic.
This is a lisp like syntax
@GDC: Is this video available on another platform? RUclips automatically blurs whole slides in this…
these kind of automatic generation its only in its infancy , further developements in artifcial neural networks will take to a new level
ANNs don't seem very well suited to this task.
+littlebigphi are you up to date with their developement?
Not really. I've just taken a machine learning class, and done some external reading. I don't know if there's anything in particular you had in mind, but I'm aware of recurrent neural networks.
Controlled generativity is in a much cruder state in neural networks. They are excellent and (mostly) unbeatable for classification and regression, with the proviso that you have a ton of data to train it. But it's not, and likely will not, be the be all end all of AI.
Even the researchers who developed the techniques that overcame the problems training large neural networks admit that they have no idea of *why* and *how* the resulting trained models do what they do.
Deep neural networks in summary. To deal with noisy data from the real world: great. To generate behaviour with any remote semblance of control over what is generated: years away, if at all possible.
Completly false ( innacurate ) neural networks excel in obtaining result in a controlled way , intituively
12:55 doves dont eat ants
They actually will if the ant insulted them first ;)
I'd love to learn more about this, can anyone recommend some sources?
Check out Emily Short I guess. There are other GDC talks on similar subjects as well
thanks!
Yikes, wish I read the comments before going to the website. I clicked the I am Human button, but hopefully that didn't do anything nasty.
What about it? I didn't see any comments about the website.
From an algorithmic / computer science pov, this is interesting.
But I will never implement something like that in my game.
Ilan Lakan Why? Please deliberate.
Because implementing the algorithm itself will take me much longer than create a decent sized hand made story dialogs of good quality
well thats looks very likely HTN planners, already they are very abstrac, just define task, and will get compund task. Maybe i will try this some day but actually i thoght its better a auto-narrative gameplay.
Dark souls dont need tell nothing you can always deduct what happened.
so you have made games in the past? which of them are worth playing for the public?
you are missing the whole point. it's about generating a lot of different stories in a procedural environment, not a single story
💌
Starting with the immersion fallacy eh? :P
Nice talk, not that great for a game tho (a Ton of time required unfortunately)
so it consuming time makes it not great?
this is such a dumb thing to say
I kept thinking about Crusader Kings 2.
I kept thinking about that somebody probably kept thinking about Crusader KIngs 2.
Man he sounds like Kermit the Frog.
Anyone else keep thinking of Crusader Kings 2?
Anyone else keep thinking of Crusader Kings 2?
Yeah, I thought about it as well