Showing posts with label robots that can learn from experience. Show all posts
Showing posts with label robots that can learn from experience. Show all posts

Wednesday, July 1, 2015

Machine ethics: The robot’s dilemma; Nature, 7/1/15

Boer Deng, Nature; Machine ethics: The robot’s dilemma:
"How ethical robots are built could have major consequences for the future of robotics, researchers say. Michael Fisher, a computer scientist at the University of Liverpool, UK, thinks that rule-bound systems could be reassuring to the public. “People are going to be scared of robots if they're not sure what it's doing,” he says. “But if we can analyse and prove the reasons for their actions, we are more likely to surmount that trust issue.” He is working with Winfield and others on a government-funded project to verify that the outcomes of ethical machine programs are always knowable.
By contrast, the machine-learning approach promises robots that can learn from experience, which could ultimately make them more flexible and useful than their more rigidly programmed counterparts. Many roboticists say that the best way forward will be a combination of approaches. “It's a bit like psychotherapy,” says Pereira. “You probably don't just use one theory.” The challenge — still unresolved — is to combine the approaches in a workable way."