koti Posted October 30, 2016 Share Posted October 30, 2016 (edited) http://www.itechpost.com/articles/47939/20161029/google-ais-encryption-better-human.htm Edited October 30, 2016 by koti 1 Link to comment Share on other sites More sharing options...
Ophiolite Posted October 30, 2016 Share Posted October 30, 2016 Come on! 1. I don't click links without having been given more insight as to the content than provided by a lengthy, but ultimately uninformative title. I believe it is good forum etiquette to provide a brief description of the topic. 2. This is a discussion forum. What did you want to discuss about the subject? 1 Link to comment Share on other sites More sharing options...
koti Posted October 30, 2016 Author Share Posted October 30, 2016 (edited) Come on! Okay, ok...I apologise for a spamy looking post, was posting from my phone and was holding my 6 month old distraction while doing it so I just pasted that title and, and... Nevermind. The article is pretty interesting. The same company which built the cryptohrapy software which the above link talks about also built an AI which beat the best human "GO" player a few months ago. They were bought by google for 500mln, Elon Musk has shares in it. I thought I'd post it here in the science news forum. I hope its ok Ophiolite? Geeez.. Edited October 30, 2016 by koti 1 Link to comment Share on other sites More sharing options...
StringJunky Posted October 30, 2016 Share Posted October 30, 2016 Okay, ok...I apologise for a spamy looking post, was posting from my phone and was holding my 6 month old distraction while doing it so I just pasted that title and, and... Nevermind. The article is pretty interesting. The same company which built the cryptohrapy software which the above link talks about also built an AI which beat the best human "GO" player a few months ago. They were bought by google for 500mln, Elon Musk has shares in it. I thought I'd post it here in the science news forum. I hope its ok Ophiolite? Geeez.. In all fairness to Ophiolite, it is a spammy title of the type we see here everyday; a sentence portion used as a thread title. 1 Link to comment Share on other sites More sharing options...
koti Posted October 30, 2016 Author Share Posted October 30, 2016 In all fairness to Ophiolite, it is a spammy title of the type we see here everyday; a sentence portion used as a thread title. Et tu, Brute? Title changed. 1 Link to comment Share on other sites More sharing options...
Ophiolite Posted October 31, 2016 Share Posted October 31, 2016 Well it wasn't up to your usual high standards Koti and that attracted my attention. And in regard to the distraction of the six month old - forget the forum and give that distraction your full attention. They will have six month distractions of their own in a surprisingly short time! 1 Link to comment Share on other sites More sharing options...
koti Posted October 31, 2016 Author Share Posted October 31, 2016 (edited) Well it wasn't up to your usual high standards Koti and that attracted my attention. And in regard to the distraction of the six month old - forget the forum and give that distraction your full attention. They will have six month distractions of their own in a surprisingly short time! Thank you Ophiolite and you're right, theres far too much spam here to have to dig through legitimate posts that look like spam. I cannot give up the forum (and other sources) completely as I'm learning too much and know too little to give up. My little, lovely distraction needs smart parents. Mom is not always on track with this notion but what can you do. Edited October 31, 2016 by koti Link to comment Share on other sites More sharing options...
Sirona Posted November 1, 2016 Share Posted November 1, 2016 (edited) This is not a better method of encryption than what we have now, but it is a very interesting project. The flaw is in assuming that although C couldn't break A and B's code, that the NSA for example with their top cryptographers couldn't break it. It would be interesting to see if the AI is able to decrypt codes written by humans. Edited November 1, 2016 by Sirona Link to comment Share on other sites More sharing options...
koti Posted November 1, 2016 Author Share Posted November 1, 2016 This is not a better method of encryption than what we have now, but it is a very interesting project. The flaw is in assuming that although C couldn't break A and B's code, that the NSA for example with their top cryptographers couldn't break it. It would be interesting to see if the AI is able to decrypt codes written by humans. Considering that these same guys built a system that beat the best GO player 4-1 on a 19x19 board it will be even more interesting to see if their AI will be able to stop humans from decrypting it's code. Link to comment Share on other sites More sharing options...
Endy0816 Posted November 1, 2016 Share Posted November 1, 2016 They might be able to come up with options we haven't, but we already have basically uncrackable encryption so long as the key remains secure. 1 Link to comment Share on other sites More sharing options...
Sirona Posted November 2, 2016 Share Posted November 2, 2016 Considering that these same guys built a system that beat the best GO player 4-1 on a 19x19 board it will be even more interesting to see if their AI will be able to stop humans from decrypting it's code. They might be able to use the data they collect to expand on human encryption/decryption also. I think it was more of a thought experiment than anything else. Link to comment Share on other sites More sharing options...
EdEarl Posted November 2, 2016 Share Posted November 2, 2016 (edited) Okay, ok...I apologise for a spamy looking post, was posting from my phone and was holding my 6 month old distraction while doing it so I just pasted that title and, and... Nevermind. The article is pretty interesting. The same company which built the cryptohrapy software which the above link talks about also built an AI which beat the best human "GO" player a few months ago. They were bought by google for 500mln, Elon Musk has shares in it. I thought I'd post it here in the science news forum. I hope its ok Ophiolite? Geeez.. According to the inventor of Deep Mind, it is a general purpose AI. In other words the same program that won Go also makes unbreakable encryption. The difference is its training for Go was millions of games (first ones on the internet, then playing against itself), and its training for encryption IDK but analogous to Go training. PS Deep Mind gets its input visually, which means that as hardware improves, smaller server farms will be required to run it, and it will be capable of training for several things without forgetting the previous training. It will need sensors for acceleration, sound, smell, touch, etc. Otherwise, it seems close to being a software solution that may become aware of itself, except the hardware isn't ready. One day, it might be trained to design a server for itself. Edited November 2, 2016 by EdEarl Link to comment Share on other sites More sharing options...
koti Posted November 2, 2016 Author Share Posted November 2, 2016 According to the inventor of Deep Mind, it is a general purpose AI. In other words the same program that won Go also makes unbreakable encryption. The difference is its training for Go was millions of games (first ones on the internet, then playing against itself), and its training for encryption IDK but analogous to Go training. I'm no expert on cryptography but this is exactly why I find the company and their AI'ish software fascinating. Considering that the amount of combinations in GO are from the practical point of view infinite, I presume that the implications of programming a piece of software to play GO very well had to be that of not using brute force but well thought out learning capabilities which is so difficult. Together with the fact that the same AI can come up with (if it's true) algorithms difficult to crack by humans this makes one hell of a breakthrough doesn't it ? Link to comment Share on other sites More sharing options...
EdEarl Posted November 2, 2016 Share Posted November 2, 2016 I'm no expert on cryptography but this is exactly why I find the company and their AI'ish software fascinating. Considering that the amount of combinations in GO are from the practical point of view infinite, I presume that the implications of programming a piece of software to play GO very well had to be that of not using brute force but well thought out learning capabilities which is so difficult. Together with the fact that the same AI can come up with (if it's true) algorithms difficult to crack by humans this makes one hell of a breakthrough doesn't it ? Yes, ... see my PS on previous post. Link to comment Share on other sites More sharing options...
koti Posted November 2, 2016 Author Share Posted November 2, 2016 Yes, ... see my PS on previous post. I see that one of your your favourite areas of interest is AI. Would you say that peoples' genuine concern about AI becoming dangerous is justified? There are some high profile people stating these concerns, Elon Musk being one of them. As I said, I'm far from having a fully informed opinion on this but it seems very far fetched or even funny to me to be concerned about a "Terminator" like scenario. Link to comment Share on other sites More sharing options...
EdEarl Posted November 2, 2016 Share Posted November 2, 2016 Musk expresses his concern as AI not being democratic; that is, under the control of a person or few people. He created OpenAI to counter that threat. We are cyborgs already, if not in fact, at least in essence, because we are dependent on computers, including cell phones. The next generation is supposed to be wearable, and several futurists envision implantable computers with links to the WWW. Whether the information on the internet makes us free or a slave will probably vary from person to person and issue to issue. There is a plethora of information on the internet that people can peruse, from news to encyclopedias. What a person knows depends on what they peruse. If you don't read the newspaper you are uninformed; if you read the newspaper you are misinformed. Mark Twain. Someone trying to understand the facts as realistically as possible may not succeed, and many people are biased. I think AI will sometimes misunderstand, especially in the beginning, and that makes them dangerous. I'm not particularly concerned about terminators. Link to comment Share on other sites More sharing options...
Ophiolite Posted November 3, 2016 Share Posted November 3, 2016 Off-topic: I see some asshole has downvoted three innocuous posts (#6,7 and 8) from three individuals, myself included. I've neutralised the other two. Link to comment Share on other sites More sharing options...
koti Posted November 3, 2016 Author Share Posted November 3, 2016 Off-topic: I see some asshole has downvoted three innocuous posts (#6,7 and 8) from three individuals, myself included. I've neutralised the other two. I noticed that someone downvoted my #7 comment and it seemed really odd as this particular comment is kinda "undownvotable" Are you sure it's a person or could it be a glitch in the matrix? Link to comment Share on other sites More sharing options...
EdEarl Posted November 3, 2016 Share Posted November 3, 2016 Off-topic: I see some asshole has downvoted three innocuous posts (#6,7 and 8) from three individuals, myself included. I've neutralised the other two. TY for your good eye and enterprise. I noticed that someone downvoted my #7 comment and it seemed really odd as this particular comment is kinda "undownvotable" Are you sure it's a person or could it be a glitch in the matrix? Perhaps it is karma. Link to comment Share on other sites More sharing options...
StringJunky Posted November 3, 2016 Share Posted November 3, 2016 I noticed that someone downvoted my #7 comment and it seemed really odd as this particular comment is kinda "undownvotable" Are you sure it's a person or could it be a glitch in the matrix? Probably somebody sulking having a little stampy-feety session 1 Link to comment Share on other sites More sharing options...
koti Posted November 3, 2016 Author Share Posted November 3, 2016 Probably somebody sulking having a little stampy-feety session Ah, a new word again. Another day has not been lost. Link to comment Share on other sites More sharing options...
Recommended Posts