Australia lags as the world tries to figure out AI boundaries – The Australian Financial Review
It is also in talks to do the same thing in the UK, although it has not been able to convince Australia’s own government to get on board.
Australia allocated $29.9 million in the 2018 budget to spend on AI research over four years, split between three organisations: CSIRO, the Department of Education and Training, and the Department of Industry. Priddis says that is a pittance by comparative global standards.
“That’s $2.5 million a year each … whereas in March last year, French President Macron announced €1.5 billion to spend in the next year on the same thing, and Germany’s Angela Merkel then announced €3 billion,” he says.
“The French economy is twice the size of ours, but the amount of money is 500 times more. These countries are comparable in terms of the sophistication of industry, government, infrastructure, education, and yet the difference could not be greater.”
For its money, Priddis says these countries are stealing a huge march on Australia, not only in gaining knowledge of the technology, but in convincing forward looking companies and leading experts to move there.
He says it is not possible to be a “fast follower” in a technology like AI that is growing in sophistication exponentially, and fears that once a country falls behind, it will only get smaller in the rear view mirror of others.
The Australian Financial Review spoke with Priddis as he was meeting with Labor MP Ed Husic, who had committed Labor to investing $4 million in a national AI centre of excellence, prior to the party’s shock 2019 federal election loss. He and Faethm are working on developing a bipartisan and cross industry group to focus on getting Australia properly AI-ready.
“There was no way [Labor’s election pledge] was the end destination of that journey, it was the start. More importantly, it would get momentum and we could have built out of that a whole lot of other work, because we need a lot more joined up effort,” Husic says.
“It’s about having a clear sense of what is needed for us to become AI ready and then backing it up seriously, so that it’s not all talk,” says Husic.
The talk itself around AI at the end of last year centred around the human input into its creation, and the ethics involved in creating systems that make increasingly important decisions.
In November, mining billionaire Andrew Forrest enlisted dozens of the world’s leading experts on artificial intelligence in what he says is an attempt to ensure the technology does good, not harm.
Then in December, the Australian Human Rights Commission released a lengthy discussion paper about the laws that need to be put in place around the use of AI, to avoid citizens’ rights being ignored.
Australia has a case study to be truly ashamed of in the misapplication of automated government processing, in the so-called ‘robodebt’ mess, where up to 700,000 cases emerged of vulnerable citizens being wrongly pursued for debts, due to faulty algorithms.
In its discussion paper, human rights commissioner Edward Santow said Australia has to get a legal grasp on technologies, which are changing lives profoundly.
“Sometimes it’s said that the world of new technology is unregulated space; that we need to dream up entirely new rules for this new era. However, our laws apply to the use of AI, as they do in every other context. The challenge is that AI can cause old problems – like unlawful discrimination – to appear in new forms,” Santow wrote.
One area of concern raised in the report was the increasing use of facial recognition software, which already performs a major role in China’s Orwellian Social Credit system, and which is being aggressively pursued by Australia’s government in the 2019 Identity-matching Services Bill.
The legislation will establish a national facial biometric matching capability, which the human rights commission fears will combine data from multiple public sources, like CCTV cameras and encrypted messages to potentially ride roughshod over rights to privacy, non-discrimination and liberty.
It advocates for a legal moratorium on the use of facial recognition technology in significant decision making until a proper legal framework is in place.
“I sometimes feel like we’re stuck in a fever dream of utopian and dystopian visions,” Santow tells the Financial Review.
“AI is ushering in genuinely revolutionary change and we can see positive and negative changes for our community, but I think we need to be cautious – not fearful but cautious. AI can make old problems – like discrimination on the base of race, gender and disability – reappear in new forms.”
The Human Rights Commission observes that many organisations, including the world’s biggest AI producing software companies like Microsoft, Google, Facebook and Salesforce have developed codes of conduct by which AI development must occur.
All have similarities and express lofty goals of ensuring that systems are built to do good for humanity and avoid entrenching unfairness.
The US Institute of Electrical and Electronics Engineers, for example, has a code of ethics incorporating eight core principles, which require software engineers to act in a manner that is in the best interests of the client and consistent with the public interest; maintain integrity and independence in their professional judgment; and advance the integrity and reputation of the profession.
However, the Commission’s report says, in reality, such policies are just stuck with a drawing pin on the backboard of a developer’s desk and largely ignored as they go about their work.
“Ethical frameworks that guide the work of professionals, particularly those published by technology companies, frequently outline a commitment to the common good. However, they often provide little practical guidance to those who design new technology products and services, or to those who purchase these products and services,” the Human Rights Commission paper says.
“Many recent ethical frameworks contain high-level value statements, which are not precisely defined. For example, the principle ‘do no harm’ is common in such frameworks, but its precise meaning in the context of developing and using new technologies is not widely understood or agreed.”
Of course the problem with agreeing to ‘do no harm’ with AI only applies to actors in the realm that are interested in sticking to the rules.
Santow says he applauds the work of Australian artificial intelligence scientist Professor Toby Walsh, who has taken a lead in global efforts to encourage the United Nations to ban AI-driven ‘killer robot’ autonomous weapons.
However, Walsh himself says he remains concerned that some countries are not committed to do the right thing by humanity and stop robots from being used to kill.
“On the positive side, an increasing number of nations have called for a pre-emptive ban on lethal autonomous weapons. We’ve got 30 nations now calling for a ban at the UN,” Walsh says.
“But on the other hand, we have an increasing arms race.”
This arms race includes Turkey’s plans to deploy it’s first “kamikaze” drones, equipped with facial recognition software, on its border with Syria. The drones can reportedly enter into a building or cave before “detecting and neutralising threats”.
The US, UK, Russia and China are all developing prototype weapons, which can kill with icy calm and on a large scale.
“The US, to its credit, has been at the forefront of military AI ethics, but there are other nations like Russia and China, where it is less clear and likely less ethical,” Walsh says.
“Facial recognition software is still struggling to pick out the differences in non-white faces, so that is just one reason why putting it in charge of a kamikaze drone is troubling and irresponsible.”
Closer to home and the major concern most Australians will still have about the expanse of AI, is its potential to kick them out of a job.
Sharon Parker, professor of organisational behaviour at Curtin University and director of the Centre for Transformative Work Design, says while individuals and governments have a role to play, most responsibility to ensure a smooth societal transition to AI-assisted employment lies in the hands of employer organisations.
She says organisations must create an environment where employees can develop the skills to make themselves useful in coming decades.
Because most workers midway through their careers cannot afford to take time out of employment to retrain, a change in approach will be required for many companies.
Without onem Australia will see a large number of professionals rendered obsolete, while its companies fret about skills shortages in newer disciplines.
“Fear [of automation] is probably not the most helpful emotion … but I don’t think being complacent would be helpful either. People do need to get ready for it and proactively try and update their skills and be willing to learn and adapt,” Parker says.
“Just about every analysis under the sun about how many jobs will be replaced by automation shows that, even if there is a whole bunch of new jobs created, they will be different.”
From a government perspective, one consideration consistently floated to avoid the breakdown of the fabric of society is the creation of a universal basic income, which would effectively amount to government funded handouts to ensure displaced workers can afford to live and keep spending.
This would be a huge call for whichever government was in office and Parker believes it is better for companies and organisations to ensure that employees have the confidence and financial means to reskill.
This means having training options in place and giving guarantees that if a worker commits to a retraining program, which effectively turns them into a non-productive business cost for a year or so, they are safe from retrenchment.
While proponents of automation have tended to push a line that AI does not replace staff, it just allows them to perform more valuable tasks instead, Parker says that in reality there are many examples of companies seeking efficiency above all else.
Too often, the early use of AI capabilities has been to replace staff or make their lives a misery, Parker says.
She cites examples of Amazon warehouse workers having their movements tracked to a level that means even the lengths of their toilet breaks are analysed by automated systems and potentially used against them, or call centre workers who are subjected to having their emotions analysed based on the tone of their voice in monitored conversations.
Then there are Uber drivers, who are lured to the platform by the promise of flexibly working on their own terms, but are then pressured by the fear of algorithmic retribution to drive at particular times and not decline jobs they don’t want to take.
“We talk about smarter technology augmenting human performance, but the reality is that the way that technology is designed and implemented in many organisations is actually making jobs worse … A focus of our research is not so much the question of will there be jobs in the future, but more the question around what sort of work will there be in the future?” Parker says.
“As humans, we need it to be high quality, meaningful, interesting, healthy and productive work. There is already a mental health crisis and yet some people see they could face the prospect of either being replaced by a robot or being treated like one.”