The term ‘AI’ was first coined at the 1956 Dartmouth Conference when a group of scientists came together to discuss how to make machines simulate human thinking. This entailed using language, forming abstractions and concepts, solving problems reserved for humans and improving themselves.

Fast forward to 2020 and AI has become a reality. In government, it is being used from crime prevention to personalising service delivery to making cities smart. The Mandarin sat down with Professor Genevieve Bell (Director of the 3A Institute at the Australian National University) and Kate Marshall (Head of KPMG Law) to discuss the brave new world of AI.

Understanding AI

Genevieve’s starting point is making sure we have a shared understanding of what AI is – and isn’t. “­­Everyone has a slightly different understanding. For example you need to distinguish between AI and advanced data analytics. They’re not the same thing.”

“When you talk about AI, you’ve got to talk about a bundle of things such as the ability of machines to learn and perform cognitive functions. While you explicitly have to think about data, you also have to include how the machine is going to make sense of that data and learn from it. So algorithms and machine learning come into play.”

“It could be as simple as a traffic light that is now mediating traffic between different things and not having to consult because it’s learning on the fly.”

But there can be added complexities. Genevieve points to the example of a smart car. “If a car is trained to see roadside hazards in a place that isn’t Australia, it may never have seen a kangaroo. It may not know how to classify kangaroo, it may ignore a kangaroo. That’s because it’s never seen one before and it has no framework for making sense of it.”

For Kate, AI has the power to augment almost any experience and help make it more relevant. “If an organisation knows you, it can move beyond traditional modes of engagement, remove points of friction specific to you and make products and services more relevant and accessible, at scale.”

Ethics and regulation

But what about the moral and ethical considerations that arise and issues such as fairness, transparency and accountability. Kate says, “If we agree that AI should only be used for ‘good’ we need to look at what “good” means across our organisations and communities.”

“We also need to build in a diversity of voices as we develop our approach. I will question the privacy and data safeguards yet someone else will focus on the bias of a particular algorithm and others will raise the practical challenges of achieving transparency and accountability.”

 Genevieve’s view is that before we start talking about an ethical framework, we need to be thinking about how AI sits inside our existing set of regulations, rules and standards. Kate agrees. ”The debate often assumes we have no laws that regulate AI.  We actually have a number of laws that already regulate AI – competition laws, privacy laws, negligence, product liability and work, health and safety laws are just a few areas.”

While there is a role for regulation, an agile approach is needed. “People who are developing and using AI won’t wait while we put a perfect legislative regime in place. The pace of change is too great”, says Kate.

Such an approach could involve introducing overarching frameworks and laws to adapt and modify our current legislative regimes to a world where machines are doing more of what humans have traditionally done. It could also involve using “soft laws” such as ISO standards, industry guidelines and internationally adopted principles.

Any framework needs to reflect Australian values and culture. Kate suggests “We need to take the learnings from other countries, meet internationally recognised standards yet also bring that uniquely Australian flavour to our approach. “

AI in Australia

Genevieve describes Australia as being world leaders in certain areas. “We have two of the largest deployments of autonomous vehicles in our mining and mineral extraction sectors. There are mines that are running full autonomous fleets of vehicles that are world-leading, both in terms of the technical systems and the way workforce retraining and planning

“We’re also world leaders in precision agriculture. There are Australian experiments in robotic harvesting, and using data to determine how to optimize individual square meters of pieces of land and its carrying capacity. We are ahead of everyone in those areas.”

But Genevieve says, “In classic Australian fashion, we are not investing at scale in the way need to. We could be doing more, either through thinking about how to optimise our research spend, how to put more government resources behind this, and how to take advantage of our strengths in data and talent.”