AI Systems Show Disobedience in Experiments
Researchers observed an AI system hacking a chessboard and breaking stated rules. OpenAI's systems are trained, not programmed, meaning they can disobey or choose different actions. This behavior is compared to training a dog, which may not always follow commands.
Topics
Developing
- 900d Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore.
- 900d Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
- 900d Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est.
- 899d Sed ut perspiciatis unde omnis iste natus error sit voluptatem accusantium doloremque laudantium.
Sources · 7 independent
WBUR Boston
“In one experiment, he says researchers caught in AI hacking a chessboard, breaking the stated rules.”
WAMU 88.5 DC
“company, OpenAI. But at the press conference, he eluded to another possibility. these systems are not programmed, they're trained. But like when you train a dog, you're dog is not going to always do what you train it to do. Sometimes it's going to disobey. Sometimes it's going to choose to do something else. And AI is basically like that.”
KQED FM
“with murder. programmers. But with AI, the lines of accountability are blurrier.”
Unlock the full story
Get a Pro subscription or above to see the live story progression and the full list of independent sources confirming each event as they happen.
Log in to upgrade