Reports of the likelihood of a robot uprising have been greatly exaggerated, according to the Director of the Australian Institute of Machine Learning, at the University of Adelaide, Professor Anton van den Hengel.
“There’s this idea that the robots are going to take over, that we face this risk of singularity, which is ridiculous. We’re years from having a robot that can even empty a dishwasher.” he says.
“If we can’t have a robot that folds socks, then the risk of there being a robot that can take over the Earth are relatively slim, I think.”
Van den Hengel has spent the last 20 years studying machine learning and AI, particularly in the field of computer visions.
Despite the limited abilities of current robots, he does see a machine learning revolution coming, based on some older ideas of futurism.
“We kind of forget that we had this dream in the ‘60s that we’d all have robot housekeepers that would understand what we want them to do. The idea then was that we’d have robots that you could tell what you want, and they could navigate the world and go and carry out a simple verbal instruction,” he says.
“One of the places we’re going with machine learning is towards a robot that might be able to take a simple instruction and actually carry it out.
“The University College London are working with Dyson to develop a robot vacuum that actually looks at your room and understands what it’s looking at, so it won’t vacuum up your socks, because it will recognise that they’re socks, not dust.”
That’s not to say that burgeoning technology isn’t dangerous, just that it’s a much more benign threat.
“Robots are very good at doing a very constrained set of operations in a very constrained space, so they’re very good at welding or sweeping the floor,” he says.
“Humans can’t go amongst the welding robots, because they’ll get killed. Quite regularly, robots do kill people who work maintaining them. They’re very dangerous things, but they’re not dangerous because they’re so intelligent they’re going to take over; they’re dangerous because they’re so daft that if you are in their space, you’ll get welded.”
The smarts behind some of our favourite technology has exploded though, with the advent of cheaper production, the inexorable amassing of data leading to algorithms that can compete, and some times exceed beyond human capabilities.
This is the swell pushing driverless cars ever closer to our shores, and even affecting the types of data-driven jobs historically left to humans, like basic accounting, and some stockbroking positions, van den Hengel says.
“The banks and the stockbrokers wouldn’t be using machine learning to make decisions if they weren’t making a hell of a lot more money out of it than they would if a human was making the decision.”
While in these early stages, such technological transformation is read widely as a threat, but van den Hengel predicts that eventually that same economic argument will be the breaking point that could even lead to society letting go of its tightly held love of the automobile.
“It won’t be long and the machines will be better at driving cars than humans are, at which point we’ll have to question whether we want humans to drive cars at all,” van den Hengel says.
“At some point, as a society, we will have to make the decision about whether we’re going to bear the cost of having thousands of people a year be injured and killed by cars, when much of that could be stopped by having driverless cars.”
Whether the fruits of the machine learning revolution will be as sugary as van den Hengel posits, it is a movement that has gripped society at large, and is influencing the way we work and live.
Van den Hengel will present on this topic at the University of Adelaide’s premier public lecture series, Research Tuesdays. The University opens its doors every month to share the latest research discoveries with the community, and is the March installment.
Van den Hengel hopes those attending will leave with a better understanding of the threats and opportunities of the coming revolution, with less of the polarising hype.
“There is a lot of hype, and a lot of fear, and most of it is unfounded, so there will be four experts on stage actually talking about the reality of what machine learning can and can’t do, and what the dangers and the opportunities actually are,” he says.
“There certainly are dangers with machine learning and artificial intelligence more broadly, as there are with any new technology. We need to decide how we’re going to deal with that, and how we can use the technology to improve people’s lives.”
Due to popularity, the Machine Rising Research Tuesdays lecture is sold out but can be live streamed on March 13 at 5:30pm via the Research Tuesday Facebook page.