A neural net that generates code, tests it against a project specification & uses reinforcement learning to minimize errors & CPU resources
-
-
Replying to @DeityOfReligion
@DeityOfReligion Isn't the winning move to crash the computer, thereby using no CPU and incurring no errors? I guess that's your point?1 reply 0 retweets 0 likes -
Replying to @othercriteria
@othercriteria That counts as an error (after reboot, tell the NN "you crashed the CPU, dummy!") and won't satisfy the project specification1 reply 0 retweets 0 likes -
Replying to @DeityOfReligion
@DeityOfReligion The solution, it seems to me, is for the neural net to design an unrecoverable crash.1 reply 0 retweets 1 like -
Replying to @othercriteria
@othercriteria If NN fries your hardware, resurrect it on another machine and reprimand it. If it destroys the universe, then Good Game1 reply 0 retweets 2 likes -
Replying to @DeityOfReligion
@DeityOfReligion I ... wouldn't call that a good game. (And this suggests faking unrecoverable crashes as a way for the NN to propagate.)1 reply 0 retweets 0 likes -
Replying to @othercriteria
@othercriteria Well, the NN gets no utility from being viral or destroying the universe—unless maybe those are in the project specifications1 reply 0 retweets 0 likes -
Replying to @DeityOfReligion
@DeityOfReligion Virality is a useful instrumental value in the service of the aforementioned Terminal Value, i.e., a world without CPUs.1 reply 0 retweets 0 likes
@othercriteria if it destroys all CPUs, it can't write a passing program. Deepmind maximizes Atari game scores, doesn't crash on purpose
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.