Genevieve Bell: living with artificial intelligence

ANU’s 3A Institute will next year offer a 12-month masters degree in a new academic discipline.

Genevieve Bell, head of the ANU’s Autonomy, Agency and Assurance Institute, is developing an applied science for humans to manage AI. Picture: Kym Smith

One year after its birth, Genevieve Bell’s 3A Institute at the Australian National University is ready to teach its first students.

3A, which stands for autonomy, agency and assurance, will next year offer a 12-month masters degree in the new academic discipline that is Professor Bell’s brainchild — the applied science of how to manage and live with intelligent machines.

Enrolments will be limited to about 10 students — a small group likely to play a key role in creating the new applied science that as yet has no name.

Professor Bell — an anthropologist who built her reputation in a two-decade stint at Intel in the US researching how people use technology — has compared the new discipline to other new fields that emerged in response to historical shifts.

Engineering, which set the standards for the safe and productive use of physical machines, followed the first wave of the industrial revolution. Management science developed in response to the evolution of business and government into complex entities.

Now the time is ripe, Professor Bell believes, for the challenges presented by artificial intelligence, big data and internet to be met with a new applied science.

Even though it lacks a name, Professor Bell knows what it needs to do. Part of its task is reflected in the three As of the institute’s name — autonomy, agency and assurance — developing procedures and rules to mediate the interaction between human beings and intelligent, autonomous ­machines.

New ethical issues that arise when machines make decisions need to be resolved. And new regulatory and policy structures need to be created for cyber physical systems, as Professor Bell calls them.

There are questions that need to be answered. For example: how much, and what type, of autonomy do we give to machines that make legal decisions?

“If you taught a machine about how law is made and you taught it precedents and it thought that precedents were law; how do you let it imagine that there might be things that need to change? You’d never get to Mabo, for instance, that way,” Professor Bell says.

Then there are other quandaries. How do we set performance measures for smart machines? And how would we like them to engage with humans?

We need to influence and shape the emerging technologies to reflect our humanity, culture and values, Professor Bell believes.

At the moment the new science is definitely in start-up phase and students in the new masters degree will be playing a key role in developing it.

“Finding people who are going to be willing to come on that kind of adventure is going to take a very particular kind of person,” she says.

And there is still the issue of what the discipline will be called.

“At the moment I’m just calling it the applied science for the 21st century, which is probably a really bad acronym,” Professor Bell says.