Not long ago, Professor Mary-Anne Williams posted a photo on her Twitter account of a robot changing a baby’s nappy. There is another amusing video on YouTube, she says with a twinkle in her voice, of a robot fetching itself a cold beer from the fridge. How marvellous.
They have a similar robot (a PR2 called GUTSY) — the only robot of its kind in Australia — in The Magic Lab which she founded at the University of Technology Sydney (UTS) 15 years ago. If they wanted a robot to do the morning coffee run and be proactive enough to pick up a few of the staff’s favourite muffins as well, her team could certainly program it.
Williams specialises in social robotics — the art of making a machine not just functionally intelligent, but emotionally and socially intelligent as well.
Her Innovation and Enterprise Research Laboratory is so cutting edge, and so well regarded internationally, that Apple co-founder Steve Wozniak and Peter Gardenfors — Sweden’s top cognitive scientist and a member of the Nobel Prize Committee for Economics — both visit Sydney to engage with her team.
“We’re not looking to commercialise some little idea,” Williams — one of an incredibly small number of women working in robotics globally — says of what attracted such international names to her UTS lab.
“We’re looking to put a dent in the Universe. We want to have a massive impact, so we’re only interested in fundamental research questions working to find insights that will enable us to pursue high-orbit innovation.”
It must feel a long way from the rural lifestyle she grew up in, as the daughter of a third generation publican in the New South Wales (NSW) town of Glen Innes. Certainly, her parents likely never imagined where her early academic ambition would lead.
“My great-grandfather owned and built 16 hotels across the country, the hotel I grew up in was built in 1906 and held for 100 years. I got into science at the University of Sydney but my parents wanted me to stay in the hotel business. My father insisted I work at the hotel and consider the life of a publican.
An avid horse rider, she decided instead to join the NSW Mounted Police. Much to her surprise, and horror, women were not allowed to apply.
“That was 1979, can you believe it? In my lifetime! But I discovered I could go to Canada and join the Mounties, so I was making all these plans until one of my cousins, who was training to be a priest, heard my father had put the kybosh on me going to university and convinced him to let me go to University of New England, as long as I came back to the hotel on Friday nights to work.
“I learned to program in Chemistry while having to compute the location of electrons around a Hydrogen atom.” But Williams went on to do a PhD in Computer Science, worked for MLC computerising the way they manage superannuation funds, and became increasingly interested in artificial intelligence.
“At the time people were thinking that chess was the benchmark artificial intelligence problem. Making a robot play chess was thought to be the pinnacle,” she says with a hint of irony. Talking to Williams, it soon becomes clear the pinnacle is far more breath-taking than that.
In August this year, the Commonwealth Bank of Australia and Stockland Group announced they would begin exploring the disruptive potential of social robots.
It’s a game-changer in Australia. ‘Chip’, as it is known, has been programmed by Williams and her team of social robotics innovators. Their job is to give Chip social intelligence, so that (unlike what she calls “factory robots” that are used in mechanised production processes) Chip can interact with and help people be more productive.
“You watch a robot get beer from the fridge and it is simply following instructions. It has no representation of what a fridge is, what people are, what beer is for — just like a Coke dispensing machine works without knowing what Coke is.
Transposing that to a home or office environment will not work, she says. They need to be able to interact, communicate, cooperate and collaborate.
That is what makes Williams work so special. “Knowing what people know is critical to communication working in any social setting,” she says. “So if robots can have some model of what people know, and their intentions, then the robot will be a lot more enjoyable to work with.
“From a research perspective it’s absolutely fascinating because we had to study people. People are proof of concept for artificial intelligence.”
Modern day robotics, she says, is the future. “When we can build a capability in a robot, we have understood and mastered it.”
Williams, who is also attached to the CODEX centre at Stanford University in the United States doing work on the ethics and issues of privacy with robotics, says it’s good to have a healthy scepticism about the potential.
“Robots collect a lot of data, so what if someone hacks your robot, can they download images of you? And if your robot is helping you in your home, should it turn away when you drop your towel to go in to the shower?”
They are unanswered questions, but it shouldn’t stop people like Williams exploring the science behind giving robots common sense reasoning, as children learn it, or to predict how people behave using programmable algorithms.
After all, wouldn’t it be nice if your robot knew that you needed a cold beer from the fridge even before you did?