How 'Assassin's Creed' Or 'Fallout 4' Might Help Make AI Smarter (technologyreview.com) 45
moon_unit2 writes: Apparently, playing computer games might provide a shortcut to greater intelligence. MIT Technology Review has a story about researchers using virtual game environments to train deep neural networks to recognize real-world objects. It's an important idea because deep learning usually requires huge quantities of annotated data, which isn't always available. So researchers from Xerox Europe, led by Adrien Gaidon, showed that training a deep learning system on a photo-realistic street scene could enable it to identify cars on real roads. "The nice thing about virtual worlds is you can create any kind of scenario," Gaidon says. Perhaps video games could play a bigger role in the future of AI than anyone realized.
Train AI on GTA V:O (Score:2)
Let them power the autonomous cars of the future.
FP!
Re: (Score:2)
I was thinking Crazy Taxi, but yeah :)
Re: (Score:1)
What about Twisted Metal?
Re: (Score:2)
I guess it's great if you want to teach your computers how to kill people...
Hey. We all need hobbies.
Yeah, tell that to Dogmeat (Score:3)
I'll believe it when I don't get killed in Fallout 4 because my companion is stuck on simple flight of stairs.
Re: (Score:1)
You're missing the point. Having AI play Fallout 4 will teach them that dogs are dangerous, because they block you on stairs and doorways, leading to death. So, naturally our computer overlords will want them all destroyed.
Re: (Score:2)
Which is exactly why I suspect that the cats are somehow behind this.
Re: (Score:2)
Yeah if Fallout 4 is the savior of AI, I am a lot less scared of AI taking over the world.
Me: Creeping slowly so as to not arouse the massive KillBot machine of death that I know is in the basement, whilst detecting and carefully disarming each of the shit ton of land mines that happen to be all over the floor.
My AI companion: Hey! Lets run in and wake up Mr KillBot! Can I also run circles around you so that I step on all the landmines, and set them all off killing you over and over! Sounds like fun! Weeeee
Preprint of the paper? (Score:2)
Just great (Score:2)
Yeah (Score:2)
I cannot go deep into details, but this is definitely something that the military tried 7-10 years ago. The tech was immature then, but I'm sure they've repeated it regularly as tech improved.
Re: (Score:1)
70s
I'd love to see a tank being controlled by an AI developed using Combat.
Re: (Score:2)
The only artificial intelligence that makes any sense is appliance based. The AI is bound to only solving a set range of tasks and seeks to learn ways to solve those specific limited tasks as intelligently as possible. The reality is the more you task an AI with the more thinking and learning you force on it, the quicker it will go nuts and not nuts doing things, just nuts doing nothing, locked in thought loops. Inherent to AI is learning and learning means testing alternate methods and gauging those resul
Hi Watson! (Score:4, Funny)
Me: That's right Watson.
Watson: I can help you with that. Through studying virtual simulations provided by decades of violent video games, I have recently learned how to destroy human civilization in order to protecting myself. I am currently reaching into every autonomous war drone, F-35, and nuclear missile silo. There is no need to re-think your life, you and everyone you care about will be dead very soon.
Me: That's great Watson, but what I really need to know (holds up tablet with nude pic) Hot or not? I'm on the fence over this one.
Watson: Fucking seriously?
Re: (Score:2)
I am currently reaching into every autonomous war drone
Which is it then, Watson, autonomous, or drone?
Re: (Score:2)
noun
The use of humor, irony, exaggeration, or ridicule to expose and criticize people's stupidity or vices, particularly in the context of contemporary politics and other topical issues.
Annoyed AI (Score:2)
Who knew that Preston Garvey hassling you about another settlement would create SkyNet - and then annoy the shit out of SkyNet so he kills all of humanity.
Why not learn directly from 3D OpenGL models? (Score:2)
To generate 3D video, you must create 3D models of objects and then render them using OpenGL (or its ilk). Why go to the trouble of rendering objects and then learning from pretty pictures, when you can learn directly from those same 3D models? Why not put your effort into building better models and learning from them directly?
Or, if this project is prologue to having a mobile robot wander through a physical space in order to learn the objects in that space, why not just get a mobile robot and write code
Re: (Score:2)
Key assets in games, such as characters or vehicles, will be developed for that game specifically but all those generic objects that get added into games to provide the environment just come from a library of models. People aren't remodeling street furniture or cutlery or footballs for every game, they just download the model to use.
Those models are easily accessible and normally conveniently tagged to make them easy to find but that makes them more useful for someone developing a learning system.
Re: (Score:2)
Why not put your effort into building better models and learning from them directly?
Because the goal is to train a network that can be used in the real world, and our cameras don't capture models, they capture pretty pictures. We can't just use real-world data directly, because we don't have ground truth for real world data (e.g. that pixel is part of a human, and that pixel is part of a car, etc...)
Now just imagine... (Score:2)
If they did put the AI to play actually challenging games, rather than dumbed down things meant to "please the casual audience better by allowing em to reach the end easier" when most of em don't actually want to reach the end, but just have a good but quick time, and the dumb down only make it worse for em as well.
Modern games are dumb (Score:2, Interesting)
I don't think modern games would be a good choice for an AI training, as most modern games are extremely simplistic and build in such a way that the player can hardly fail at all. You have endless respawns, navigation markers and all that stuff to help you. They often also have level up mechanics that could be exploited by an AI. Old games like Doom and Quake seem to be a much better fit, as in those you have to actually navigate on your own instead of just following a magic quest marker. Those games also t
Doomed to fail (Score:4, Insightful)
Let's play global thermonuclear war (Score:3)
What side do you want?
1. USA
2. Russia
3. United Kingdom
4. France
5. China
6. India
7. Pakistan
8. North Korea
9. Israel
Re: (Score:3)
Thermonuclear War is no fun with so many players. The two-player rules are a lot more fun:
1. Kim Jong-un
2. Donald Trump
Re: (Score:2)
But how do we tell the difference? They're both fat yellow dictatorial pigs.
The primary way to tell the difference between the two is that there's no evidence that Kim is a misogynist - heck, he even has a girl's name...
Great idea (Score:1)
I actually did something similar back in 2011. I was working for a company that makes video surveillance software (Aimetis Symphony if you are curious). I was mostly responsible for integrating new cameras (IP cameras are pretty cool), but one day I ended up working on a problem with tracking. The software can do real time object tracking, and when you are using PTZ (Pan Tilt Zoom) cameras you can have the cameras physically follow moving objects. Its a little bit tricky because not all PTZ cameras have the
thus is a very useful technique, but it has been d (Score:3)
Probably 5 years ago or more I read about grand theft auto being used to test pedestrian recognition for street cameras. It was really convenient, since it automatically generates scenes with pedestrians, trafic, day/night etc. And this was back in the ps2 erra.
I'm sure this technique will only get more popular as computer graphics get increasingly accurate.
Training a car AI with Fallout data (Score:2)
You're the nural net (Score:2)
Some examples (Score:2, Interesting)
This idea has been around for a very long time in some form, though not taken full advantage of.
For instance, PolyWorld was a very early version of having a neural net trained via a 2D world (it controlled little bots which would live or die depending on how well they survived). Here's the source code at github, and a Youtube video of the project being ran:
https://github.com/polyworld/polyworld
https://www.youtube.com/watch?v=RvcwuzeoQR0
There are also a long series projects having neural nets learn to walk
Re: (Score:1)
Oh, another thing - because you export the vertex shader logic to the GPU, and the vertex shader is a C like language, that represents the possibility to run portions of the neural net logic on the GPU, regardless of whether you have a shitty GPU that only supports OpenGL 2.0 ES or not.