Last week, MartÃn Abadi and David G. Andersen with Google's Google Brain project announced that 2 of their neural network AIs (Alice & Bob) had generated their own encryption method, allowing them to hide messages between them from a third neural network AI (Eve), that was attempting read the message.
Abadi and Andersen, set this experiment up, by assigning each neural networks with specific tasks. Alice was instructed to pass a secret message to Bob, and Eve was tasked with figuring out how to intercept and read the message. After, about 15,000 attempts, Alice and Bob were able to simultaneously start using the same encryption method. In the end Alice was able to encrypt a 16 bit message, without Eve being able to fully decrypt the message. The actions of these 2 neural networks are no complicated that the researchers are unable to explain how it happened, or by what encryption method was used. It seems to me that we may need to wondering if androids dream of electric sheep.
Thanks for stopping by again. I appreciate your support. Make sure you keep checking in for more new content. So what do you think? Is AI good or bad thing for mankind? Are we an evolutionary dead end working toward our own demise? Make sure to let us know what you think. Don't forget to follow @BaicunnPress on Twitter and join the BaicunnPress Nation over on Facebook at @BaicunnPress. Also don't forget that this Saturday (11/5/16) is Extra Life. BaiclunnPress is teaming up with @RUM_Gaming to raise money for sick kids and the Children's Miracle Network Hospitals. We've taken down some adds on our site to make way links to donate. Be safe, have fun, and don't quit your job, you're halfway done with this week. (I'm probably projecting)
-Chali Baicunn

No comments:
Post a Comment
Comment & Share