Resistance may not be futile with our robot overlords

Wednesday, April 26th, 2017

TED Talks delves into ways automation can ?augment? careers and everyday life

The Vancouver Sun

Whether we should fear future “robotic overlords,” job-killing technologies born out of artificial intelligence and automation, will really depend on who is building them and what they’re building them for, according to AI developer Tom Gruber.

Gruber, co-creator of Siri, the voice-command AI feature on Apple iPhones, told the audience at TED Talks that we have a choice to create “humanistic AI” that can enhance human abilities and help people solve problems, not simply take over from them.

“We can choose AI to automate and compete with us,” Gruber said. “Or we can use AI to augment and collaborate with us to overcome our cognitive limitations, so we can do what we want to do and do it better.”

TED Talks, underway this week in Vancouver, tackled the topic of just what artificial intelligence and robotics can achieve for society and how to address problems that might come with that, such as increased unemployment, across a couple of its sessions Tuesday.

Some of the ideas that emerged from the 12 discussions held out incredible promise for innovation, but they’ll require new ways of thinking about work and income distribution if computers and machines take over a lot of jobs that people do now.

Gruber’s talk, in the session titled Robotic Overlords, focused on how artificial intelligence is being used to bridge communication barriers for people with disabilities and augment the abilities of professionals with object-recognition technology.

He used the story of researchers who used machine learning to devise a computer program that could recognize cancer cells in tissue samples as an example of how AI is helping improve the accuracy of medical diagnoses.

The experiment, Gruber said, found that the human was still better than the AI program, but when using them together, it eliminated a lot of errors and improved the hospital’s diagnosis accuracy to 99.5 per cent.

“The lesson here is that by combining the abilities of a human and a machine, we’ve created a partnership that had super-human performance,” Gruber said.

The session had other examples, such as Harvard robotics engineer Radhika Nagpal’s research into self-organizing “swarms” of robots to the advances in near-autonomous machines being designed by the firm Boston Dynamics.

“I’m not afraid at all,” said Boston Dynamics CEO Marc Raibert about the prospect of his robots taking over jobs from humans.

Robots can do dangerous tasks humans shouldn’t, Raibert said, such as helping clean up from the Fukushima nuclear disaster in Japan. Or they can be used for jobs we worry that there won’t be enough humans to take on, such as in helping look after the physical needs of an aging population.

However, the capabilities of AI and robotics have begun encroaching on tasks that used to be intrinsically human, said Martin Ford, the futurist who wrote the book Rise of the Robots: Technology and the Threat of a Jobless Future.

Accountants, journalists and even doctors are seeing at least parts of their work being taken up by automation, Ford said in his talk during the session titled The Human Response.

“We have to find a way to decouple incomes from traditional work,” Ford said and the best way he’s seen is to implement some form of guaranteed basic income as a place to start. He estimates doing so will become imperative.

Providing people with basic guaranteed incomes is also a better bet for alleviating poverty, said Dutch historian Rutger Bregman in his talk during the same session.

A four-year experiment with basic incomes in Dauphin, Man., in the early 1970s showed the concept’s success, said Bregman.

School achievement improved, hospitalization rates fell, domestic violence declined and people didn’t quit jobs because they had the guarantee of income, Bregman said.

“I believe basic income would work like venture capital for people,” Bregman said.

© 2017 Postmedia Network Inc

Comments are closed.