AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
1–10 of 47 posts
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#2> It’s not the most powerful or widely used form of AI at the moment, but it is making something of a comeback. The ability to crack Q*bert could be read as a good omen that evolutionary algorithms are going to be very useful in the future.
Wow that's quite a jump to make
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#3The case is an example of wireheading [1] and illustrates the difficulty of eliciting behaviors we actually desire from complex systems we do not fully understand.
[1] https://wiki.lesswrong.com/wiki/Wireheading
Another lesson: Evolutionary algorithms are really hard to control. Using neural networks developed through evolutionary algorithms means that we are employing a mostly opaque (though not entirely black) box created by a mechanism we can't mentally keep track of in detail. Hope that they are not deployed to control any critical systems until we get a much better grasp of them.
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#4> It’s not the most powerful or widely used form of AI at the moment, but it is making something of a comeback. The ability to crack Q*bert could be read as a good omen that evolutionary algorithms are going to be very useful in the future. Wow that's quite a jump to make
This sound like me at the end of every school essay. A forced and over-broad conclusion just to get a "proper" ending.
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#5The case is an example of wireheading [1] and illustrates the difficulty of eliciting behaviors we actually desire from complex systems we do not fully understand. [1] https://wiki.lesswrong.com/wiki/Wireheading Another lesson: Evolutionary algorithms are really hard to control. Using neural networks developed through evolutionary algorithms means that we are employing a mostly opaque (though not entirely black) box…
Has anyone been able to comprehensively state all of essential human values for a general AI to follow? Thankfully, we do not yet have an operational AGI and it is still quite a bit away from reality. (Narrow AIs we are using do not pose much of a problem because they are limited in capabilities.)
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#6Well how do you say what's cheating or not? It works and it increases the evaluation score
In this case one possible workaround to "cheating" would be to reduce the control precision, add some jittering to control inputs or change the goal function. But I'd say if it's being done solely with using the intended controls it's not cheating (as opposed to changing memory or using a debug 'cheat code').
Still, even in real sports some "cheating" is allowed (see Fosbury Flop)
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#7> It’s not the most powerful or widely used form of AI at the moment, but it is making something of a comeback. The ability to crack Q*bert could be read as a good omen that evolutionary algorithms are going to be very useful in the future. Wow that's quite a jump to make
This sound like me at the end of every school essay. A forced and over-broad conclusion just to get a "proper" ending.
@#$&%!
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#8I always found this a good project to demonstrate AI :https://xviniette.github.io/FlappyLearning/ ( based on Neuro evolution ) - speed it up for faster results
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#9So, it's basically working as a goal-oriented fuzzer.
Re: AI Cheats at Old Atari Games by Finding Unknown Bugs in the Code
#10This reminds me of AI research using NES Games. The AI eventually became proficient at completing Mario levels, and along the way it discovered novel strategies for survival, obtaining points, and finishing levels.
> Check out this timestamp to watch the machine "cheat": https://youtu.be/xOCurBYI_gY?t=9m55s
> Researcher's site about the project: http://www.cs.cmu.edu/~tom7/mario/
> The Paper: The First Level of Super Mario Bros. is Easy with Lexicographic Orderings and Time Travel...after that it gets a little tricky.: http://www.cs.cmu.edu/~tom7/mario/mario.pdf