It is not clear whether pessimism or optimism over artificial intelligence (AI) is winning the day, according to a panel discussing the state of the field at the World Economic Forum in Davos, Switzerland, yesterday.
“We’re going to keep seeing things that we thought relied on human ingenuity fall more and more into the hands of AI,” said Andrew Moore, dean of the school of computer science at Carnegie Mellon University.
Yet he is optimistic. “Many of us see a safer world because of AI,” he said.
AI systems have traditionally been good at unemotional, problem-solving tasks such as playing chess. But now the emotional understanding is growing, said Moore. This pens up the possibility of AI being used in a wider range of fields, such as health.
Taking over the world
Robots will make many jobs redundant, the panel predicted. “For now, AI is claiming gradual success in removing the boring parts of white collar work,” said Moore. However, he foresees “quite terrifying changes” in the jobs market, with AI shrinking the legal and medical professions.
A WEF report released this week predicts the loss of 7.1 million jobs over the next five years in 15 leading economies, while helping create just 2 million new jobs over the same period.
According to Stuart Russell, professor of computer science and founder of the Center for Intelligent Systems at University of California, Berkeley, 2015 was a glimpse of the future. “It was the first time you could take your hands off the wheel of the car,” he said.
He has concerns that AI will eventually take over the world. “Last year the Deepmind start-up demonstrated a learning system that resembles a new-born baby. Engineers exposed a new AI algorithm to Atari video games. The AI knows nothing about the game before it begins. It knows nothing about anything, whether it’s space or time. Within a few hours it can play most of the games at a super-human level,” said Russell.
“If your baby could play games at a super human level after just one afternoon, you might be a little concerned.”
Last year Russell was the first signatory of an open letter which asked for researchers’ help in “keeping artificial intelligence beneficial”. Other signatories on the letter are Stephen Hawking, who says AI could spell the end of the human race, and entrepreneur Elon Musk, who says AI is "like summoning the demon".
“We need to start doing research now to begin thinking about controlling AI,” said Russell. “If we knew a giant asteroid was going to crash into the earth in 75 years’ time, we would start working on the problem right away.”
The opportunities of AI outweigh the threats, said Matthew Grob, chief technology officer at Qualcomm. “But we have to be mindful of security. Yet even there, you can use AI to make improvements. An AI agent can recognise unusual behaviour on your phones, for example.”
If nothing else, AI threatens to make us lazy. “As we get more dependent on machines, human beings will become less intelligent,” predicted Ya-Qin Zhang president of Chinese search engine giant Baidu. “I mean before long, we might not know how to drive. That’s a concern to have.”
Today, search engines are good at processing or indexing but can understand little about the questions asked of them.
Apple’s voice assistant, Siri, is cute but really does not understand what a person is saying – it has a small library of prepared answers, said Russell. The Watson supercomputer can beat anyone at Mastermind but it would not get very far at a poker table.
The self-driving Google car gets confused if it confronts a cyclist approaching the wrong direction– it is not sure if it is on the right side, or if the cyclist is.“Rules tell the autonomous car what to do but not why it should do them,” said Russell. “The car doesn’t know that people don’t like being crashed into.”