Despite Elon Musk’s dire warning that thinking machines may seek to eradicate humanity, Hannah Devlin reports in The Guardian that most computer experts think “killer robots” are the least of our worries. Computer scientist Nigel Shadbolt of University of Oxford tells Devlin that “if there [are] killer robots, it will be because we’ve been stupid enough to give it the instructions or software for it to do that without having a human in the loop deciding.” Shadbolt predicts that intelligent robots will largely benefit humanity in everything from medicine to manufacturing, but he says we urgently need to have an ethical debate about “how AI is applied, including the need for transparency around how machines make decisions and of how personal data, including medical records, are used.” Here’s an excerpt from the article:
Similarly, Shadbolt suggested, the prospect of humans developing emotional bonds with machines was not entirely uncharted territory. “We’ll embue [robots] with lots of human qualities, we will begin to empathise with them” he said. “That doesn’t require these systems to be self-aware. You anthropomorphise your goldfish at home. I certainly did that with a teddy bear when I was a child.”
However, he acknowledged that latest advances in AI, which include the ability to not only interpret photos and videos but to artificially generate this material, raised “uncomfortable” new possibilities.
“A bereaved widow [could] decide to keep her husband’s voice around on her Alexa,” he said. “There’ll be the digital posthumous voice and character of a loved one. That is going to happen. These systems won’t just be faded photos they’ll be capable of creating new conversations.”
Image via Business Insider.