© 2020 Relevant Protocols Inc.
© 2020 Relevant Protocols Inc.
Relevant
Hot
New
Spam
Relevant
Hot
New
Spam
0
2.9K
0
2.9K
>"By teaching machines to understand our true desires, one scientist hopes to avoid the potentially disastrous consequences of having them do what we command."
>"By teaching machines to understand our true desires, one scientist hopes to avoid the potentially disastrous consequences of having them do what we command."
[@crosslucid](/user/profile/crosslucid) [@taylore](/user/profile/taylore) This is an excellent, as well as frightening, article, making very clear the impossibility of achieving human-like reasoning capabilities in computing machines. Neither symbolic programming nor neural programming can ever come close to capturing the immense complexity and subtlety of how the human brain works. The "law of unintended consequences," along with the propensity of humans to make mistakes underlies every software bug. There are too many instances of fatal or near-fatal errors, even in carefully-supervised systems, from the Mariner 1 to the Mars Climate Orbiter to the Ariane 5 [#501](/altwoke/new/501) to the Patriot Missile Error (Saudi Arabia, 1991), only to name a few. Blood has been spilled and treasure lost due to programming errors. The spectre of "Skynet" in "Terminator" is not an unrealistic outcome. Caveat emptor.
[@crosslucid](/user/profile/crosslucid) [@taylore](/user/profile/taylore) This is an excellent, as well as frightening, article, making very clear the impossibility of achieving human-like reasoning capabilities in computing machines. Neither symbolic programming nor neural programming can ever come close to capturing the immense complexity and subtlety of how the human brain works. The "law of unintended consequences," along with the propensity of humans to make mistakes underlies every software bug. There are too many instances of fatal or near-fatal errors, even in carefully-supervised systems, from the Mariner 1 to the Mars Climate Orbiter to the Ariane 5 [#501](/altwoke/new/501) to the Patriot Missile Error (Saudi Arabia, 1991), only to name a few. Blood has been spilled and treasure lost due to programming errors. The spectre of "Skynet" in "Terminator" is not an unrealistic outcome. Caveat emptor.
Yes artificial intelligence is one of the best technology which I ever had seen as we can train any robot car as per of our desire,one of the beautiful example of that is cozmo robot. Slowly it will be improved a lot because in this technology machines learns from the surrounding environment. But on the other side yes it's also true that by teaching machine what our desire is,also very dangerous because they lack nuance, they can do anything to fulfil our desire without giving warning.
Yes artificial intelligence is one of the best technology which I ever had seen as we can train any robot car as per of our desire,one of the beautiful example of that is cozmo robot. Slowly it will be improved a lot because in this technology machines learns from the surrounding environment. But on the other side yes it's also true that by teaching machine what our desire is,also very dangerous because they lack nuance, they can do anything to fulfil our desire without giving warning.
Some low-ranking comments may have been hidden.
Some low-ranking comments may have been hidden.