here is link to the article. https://www.theverge.com/2019/3/6/18222203/video-game-ai-future-procedural-generation-deep-learning
The guys over at Unity are banking heavily on machine learning as the engine’s primary AI creation feature (to the point where more traditional and more statistical functions and features are ignored). However, a number of users are not on-board with that and there’s actually some potentially major downsides for using deep learning if you are an indie or a small studio. Disclaimer, I don’t have actual research backing this up, but part of it is based on the current usage in image denoising.
- AI based on machine learning requires training, lots of training. That means you would have to make your bad guys playable characters for this purpose, and for proper training might need a second person to act as the player.
- AI training has to be largely redone if you add, remove, or noticeably change the variables where the training takes place. If you added new moves to your game AI, new items for them to use, and new level structures for them to deal with, it’s back to the drawing board.
- You have to be careful to not run into the problem known as “over fitting”
- You don’t have code that can quickly and easily be extended and fixed as you test your game.
If you run a large studio and can afford having a small team of people constantly playing builds to change and update the training data, then it could indeed be a revolutionary concept, but most devs. do not have that ability.
open artificial intelligence has been working on unsupervised learning.making artificial intelligence curious.unsupervised learning does not require all that training data.
An ongoing problem with AI at this point is determining what the “training” should be based on – what the model should, and should not, “see.” (And, how do you know.) We have had great(!) success with some things, such as speech recognition (thank you, Dr. Kurzweil …), but not nearly so much e.g. with character recognition. In the context of a video game where you wanted to have “truly smart” bad guys and such, changes to the scenario which would be “simple and obvious” to a human brain might utterly disrupt the model’s behavior. AI is no longer “in its infancy,” but it’s still a very little kid. I don’t think it’s really walking yet, but it shows great promise.
from the article:
…If you asked video game fans what an idealized, not-yet-possible piece of interactive entertainment might look like in 10 or even 20 years from now, they might describe something eerily similar to the software featured in Orson Scott Card’s sci-fi classic Ender’s Game . In his novel, Card imagined a military-grade simulation anchored by an advanced, inscrutable artificial intelligence…
Clearly a literary-trained individual is trying to sell a marketing article based on things she doesn’t understand.
tl;dr; it’s all marketing. You can sleep safely.
should read about a lot of the stuff they have done with artificial intelligence on ai reddit .com
I problem with using the AI machine learning software, is the large database. I don’t know if it would be apporpriate on anything except on a companies server.
look up deep reinforced learning.
Here is a bit of game creation through AI:
When I came across this video last year, my initial thought was that it was just some fancy AI hype. Curated to work in some special corner cases.
But the guy behind it (Andrew Maximov) is set to speak at the next Total Chaos conference in May. And last month he was featured in one of Chaos Groups CG Garage Podcasts.
So maybe there are some real substance behind this tool.
I hurd about that too.
too bad google stadia would not benefit the upbge by increasing the framerate of the videogames we make.
I’m not so sure …
google has created Deep Mind …
and now Stadia …
if I am diabolical I make associations …
imagine the staff Google … what is preparing for the future …
I sure hope so.
what if people could upload there videogames made with with the bge or upbge. to ai websight and the ai could recode the game then make it compatible with another game engine with better framerate.the videogame would look the same but have better framerate.
apparently the games worrld are about to change quickly …
Nvidia and Amd for examples… wants to radically change its market … and probably in view of how in the future the games will be created and used
It won’t change quickly with Stadia (lol sounds like something in a BioShock game). Most internet service is still shit (at least in the US) and will not improve anytime soon.
I refuse to participate in this hipster version a dumb terminal connected to a mainframe. Not even a master level necromancer will be able to bring OnLive back from the dead.
so if while you are browsing on intenrnet, it happens that they propose you to try an AAA game, maybe for free with just one click, as if you were playing a video, wouldn’t you do it because it’s a shit proposal?
if it is as you say, that the connection will not allow to play well on these videogames on streaming, the first to lose credibility will be themselves …
yes that would be bad for their business.
There’s also data-caps, our ISP introduced caps of nearly 1 terabyte of data to reduce the bandwidth pressure caused by online gamers and streaming gamers.
Game streaming uses a ton of data if you want decent quality, and is one reason why unlimited data plans have nearly gone extinct.