Are the robots coming for us?
“AI is currently being developed in a useful way, on balance,” he says.
“There will be job losses, but I think society will adapt and new opportunities will emerge, so there’s no need to worry about this.”
Kaila Colbin, the New Zealand ambassador of Silicon Valley thinktank Singularity University, nonetheless says the issue demands our close attention.
“I have study after study saying we should be terrified, and as many saying new jobs will be created and there’s nothing to worry about.”
The main point of disagreement, she says, is not whether jobs will go away — but whether new ones will be created.
And if that is the case, at best, we are facing a major transition, and at worst, mass unemployment.
“Meanwhile, a report from the Roosevelt Institute just came out that said a Universal Basic Income would add $ 2.5 trillion to the US economy.
“So why wouldn’t we be starting to consider our options now?” Whatever the case, most Kiwis don’t appear to be fretting.
A Massey University study found 87.5 per cent of respondents either disagreed or strongly disagreed with the statement “smart technology, artificial intelligence, robotics or algorithms could take my job”.
But Yeap and Colbin — both speaking at Great Barrier Island’s Battle of the Brains festival this weekend — agree the pace of innovation has become rapid, particularly because of developments in robotics, big data and the internet.
We’re seeing extraordinary advances in image recognition, which self- driving cars need to identify a pedestrian, tree or stop sign.
Speech recognition is being used for real- time translation — you can Skype with people who speak a different language and Skype will translate for you on the fly.
And big data analysis is leading to incredible discoveries about genetics, as well as powering something as mundane as your Google search results.
How we use robots is limited only by our imagination, says Yeap, who has tracked the advent of bipedal, human- like androids. We’ll see them becoming more emotionally capable and providing critical services, such as performing surgery.
Perhaps the biggest barrier to achieving “true AI” lies in the way in which researchers are trained to develop AI software, Yeap says.
“The whole field is focused on performance- based modelling, and if and when they try to model how the mind works, they will face a mindmodelling conundrum.
“For instance, how could they develop models of the mind when they have no idea what the mind computes?”
Researchers working in the cognitive sciences have been trying to figure out how the mind works for centuries — and they are still in the dark.
Without this knowledge, AI researchers are at a loss in building true AI and, consequently, many don’t even attempt to do so.
“Overlooking this challenge and focussing simply on performancebased modelling is a huge mistake, and it’s a trend that saddens me,” Yeap says.
“To date, the lessons humans have learned through evolution have not been incorporated into AI — and by not paying attention to the mind, the industry risks creating dumb but powerful machines.”
So what can we expect in the shortterm?
Yeap predicts big advances in the development of humanoids — twolegged, two- handed, human- like robots — and perhaps even artificial soldiers programmed to kill.
As scary as that might sound, Colbin says the real concern isn’t with robots rebelling, but simply ignoring us in pursuit of their prime directive.
An example of this control problem could be AI programmed to manufacture paper clips, becoming super intelligent, and ultimately using all of its resources to optimise production and potentially turning the entire planet into paper clips.
“The good news is that lots of very intelligent people recognise that this is the real problem, and are working hard to make sure we don’t all turn into paper clips.”