By moving aircraft into powerful jetstreams – fast-flowing air currents – Qantas expects the system will cut its annual fuel bill by $40 million and reduce carbon dioxide emissions by 50 million kilograms. It could also help it manage its ambitious Project Sunrise to fly customers directly to London in 20 hours.
While Australia’s national carrier and some other leading companies are at the forefront of applying “narrow” AI (the application of “general” AI, where machines think and reason like humans, is still probably decades away), many companies are looking on agog at the rapid pace of technological change and remain confused about where to start.
Only 9 per cent of Australian companies are considered pioneers of artificial intelligence technology, according to a study by local AI start-up Daisee, which is less than half the number in the United States. A further 12 per cent of Australian companies are experimenting, and one in three are investigating AI, but many remain paralysed by the fear of disruption and workforce upheaval.
But Joyce says a more positive mindset about jobs is required to ensure Australia unlocks the new technology’s vast potential. “I know people are scared automation can replace jobs, but our experience is it creates jobs,” he says.
With artificial intelligence proliferating in consumer devices such as voice-controlled assistants and smartphone applications with AI systems being rolled out across global industries, from diagnosing disease to allocating credit, the issue of AI exploded at The Australian Financial Review Business Summit last week, where Harvard University professor Niall Ferguson reminded delegates the technology is now of geopolitical significance.
Global spending on AI will increase from $12 billion in 2017 to $56 billion in 2021, according to Deloitte, as the number of machine learning pilots and implementations double from this year to 2020. Management consultants are calling out incredible efficiency gains that could be unlocked: the McKinsey Global Institute reckons AI could potentially deliver additional economic output of about $13 trillion by 2030, boosting global gross domestic product by about 1.2 per cent a year.
In order to gauge Australia’s preparedness for this emerging future, AFR Weekend conducted a survey of CEOs of the largest ASX-listed companies during the recent interim profit reporting season and found a variety of AI applications across a diverse range of industries. These can provide a road map for companies wanting to explore how AI might work for them.
Along with Qantas, another advanced thinker is engineering group WorleyParsons. It has created Advisian Digital to co-ordinate a new, data-led strategy with artificial intelligence systems at its heart, which the company is using to reduce risks of workplace injuries.
“It’s the largest area of investment that we have in our business,” says CEO Andrew Wood.
“We’re investing millions of dollars in AI and automation, in data analytics, in things as varied as predictive safety.
“We’ve launched a new product on the market called SaltGrid, which looks at historical data of safety performance on a site, and it can predict where there potentially are risks for activities you’re choosing to do the next day or the next week. We’ve started to roll out that product around the world, and that’s artificial intelligence.”
The SaltGrid product is helping WorleyParsons overcome a particular problem: incidents of serious injury had not been reducing. By feeding into the system a decade of OH&S incident data executives and front-line supervisors are being guided about where to intervene.
“Business managers can be good at understanding a single risk and mitigating that, but we are poor at understanding how risks are shifting in relation to others and how one is creating a causality chain,” Bradley Andrews, the president of Advisian Digital, tells AFR Weekend from his base in Houston, Texas.
“AI makes the calculations so quick – it sees something before everybody else does.”
As new data is gathered from a range of web-connected devices, the company plans to move things into real-time, so a worker could be buzzed by a wearable device to tell them that risk conditions are elevated and they need to reassess their personal safety or vacate an area.
WorleyParsons is also using AI to identify methane leaks, by putting gas quantification algorithms onto infrared cameras, and is working in partnership with CSIRO on machines that can decide whether ore is good or bad and thus worth mining.
Manufacturing, retail users
AFR Weekend identified several other leading Australian companies investing in AI. The manufacturing sector is an early adopter, where the technology is being used to assist with maintenance and spot product defects.
Cochlear bought the rights to AI software 18 months ago, using it to enable more streamlined programming of its bionic ear implant devices. Trials over the past 12 months have delivered “very encouraging” results, says CEO Dig Howitt.
“Artificial intelligence and connectivity are two opportunities for us to automate the routine checkups that are done on the performance of the processor in the implant, and to reduce the lifetime costs of care, which enables more people to get access and takes the pressure off health budgets.”
Recruitment is another sector making use of the ability of machines to read CVs. Seek CEO Andrew Bassat says the executive search firm is using AI to capture data about the people and job ads to “let us do much smarter search”.
“Ultimately you should be able to use this to help the hirer make better selection decisions based on the fact we know which candidates are likely, more likely than not, to be, successful in the individual roles. There’s a huge area we can get better and better at by using data and AI,” Bassat says.
Financial services companies are also experimenting using AI systems to read loan applications, while bigger data sets will also unleash AI systems for making credit decisions and to aid regulator compliance. ANZ Banking Group has begun to use it to guide the direction of internal auditing resources, for example the monitoring of staff expense reporting.
The retail sector is also getting on board, to help both customer service and to streamline supply chains. Coles says it is using AI to improve customer communication with more targeted, personalised and effective offers. It also has systems forecasting consumer demand, supporting better replenishment and inventory management.
Afterpay Touch is using AI to help merchants with stock management and delivery. “We are utilising our data sets to help our customers and our retailers to both better understand the customer base and for customers to get more insights into what offers are out there,” says its group head, David Hancock. “It is an ongoing commitment and one of the key tenets of a platform is how do we connect customers and merchants together and AI and data is critical to that and making the platform really hum.”
The energy sector is active in AI investment. Power utilities are looking to AI to stabilise the grid as renewables are fed into the system and to conduct predictive analytics for equipment failure. Resources companies have been using AI to control automated fleets of vehicles and processing plants.
Even the regulators are jumping on board. The Australian Securities and Investments Commission has begun trials of natural language processing technology, a sub-field of AI, to detect problematic financial advice, misleading internet advertisements for financial products and to monitor insurance sales teams’ calls. It says this will help it regulate the advice sector in the wake of the banking royal commission.
“AI now provides that new hybrid of labour and capital that is going to augment your humans and amplify your existing legacy systems. This is where we are going to see the growth,” says Catriona Wallace, the CEO of AI start-up Flamingo.
Companies are being urged to consider how to put the technology in the hands of their workers to improve productivity rather than getting distracted with how to use it to fully automate certain tasks. The Australian Institute of Machine Learning at the University of Adelaide told last year’s Senate Select Committee on the Future of Work and Workers that while AI “is the subject of much, and sometimes fantastical, speculation, when companies who have adopted AI are surveyed, they report that AI delivers improvements in sales and efficiency, and results in new jobs and new roles within companies”.
As artificial intelligence systems are rolled out at the top end of town, many smaller companies are unprepared for this technological tsunami and are being warned to consider carefully the inherent risks involved in mining data.
On a Monday afternoon this month, at the Australian Institute of Company Directors annual conference in Sydney, more than 200 packed into a standing-room-only breakout session on AI, and questions from the floor revealed widespread uncertainty about where to begin.
Maile Carnegie, the former head of Google in Australia and ANZ’s group executive for digital and transformation, said the quality and governance of data, the raw material that is fed into the AI engines, is the key issue holding back broader applications. “My assumption is the quality of your data is not where it needs to be, and it can be very expensive to get data in sufficient shape so it can actually be used,” she said. “It doesn’t matter how many use cases there are, if you don’t get your data in shape that is always going to be your Achilles heel.”
Lee Hickin, Microsoft’s national technology officer for Australia, told directors that using AI systems would require a change of mindset in the boardroom. “AI is about probablistic outcomes,” he said. “The data will not give you the absolute solutions, it will give you a sense of what is going on.”
He agreed with Carnegie that the quality of data was paramount, warning that companies that used data with inherent bias baked in were likely to trip up when AI systems analysed it.
“The more data you have, the more ‘intelligent’ your AI is going to be, but what we tend to be doing today is feeding the AI services of today with the data of yesterday. And the data we have been capturing in many of our industries is inherently biased because of the way we have thought about issues and data collection,” Hickin said.
“Just because you have data, don’t assume that is going to produce a strong AI output that is unbiased.”
The lack of diversity built into computer code is a huge problem for companies wanting to turn on AI, says Flamingo’s Wallace. “A really big thing to think about is human rights and gender and other bias in the machine.”
She also warns about data security. “Security is a massive challenge. We will see some big security breaches involving AI in the next 12 months. There will be some big blow-ups coming.”
One leading company trying to navigate others through the artificial intelligence maze is ASX.
It has built a big-data platform that soon will be offered to external companies to help them clean, curate, govern and control data sets in a highly secure environment. It will provide users with AI tools to allow them to gain insights from crunching data. Like the global cloud computing providers Amazon Web Services, Microsoft and Google, ASX is seeking to democratise AI by reducing the barrier to entry for financial market users.
“We have created a platform with ASX’s data and made it an open infrastructure so they can bring their data and use ASX’s tools. When you bring multiple data sets is when it becomes interesting,” says CEO Dominic Stevens. “Users can chose an AI or machine learning tool and go off and do something with it.”
At the AICD event, there was much discussion about whether to use AI services provided by outsiders or start-ups, or whether these should be controlled in-house. Carnegie warned that any company wanting to use AI would want to have significant resources in-house.
“The future brain of your organisation is going to be AI-enabled, so completely outsourcing and lobotomising yourself and asking other people to be the brain of your organisation doesn’t make a lot of sense,” she said. “You should be thinking about how you are going to build that brain yourself rather than being forever wedded to a big or small vendor to be the cognitive thinking processing part of your business.”
Companies using AI will also be subjected to new regulation and ethical frameworks to be put into place.
As AI becomes more prevalent – and as computers become more capable of gaining their own insights and making more autonomous decisions based on their learning – public perceptions that AI could spin out of control are likely to grow. This will mean companies need to ensure that the reasons that particular decisions are made must be explained by the systems themselves.
“When companies use AI decision-making systems, they must build them in a way that allows a person to understand the basis of decisions that affect them. This is fundamental to ensuring accountability and will be really important for all companies that use AI,” says Ed Santow, human rights commissioner at the Australian Human Rights Commission, which has called for the establishment of an “AI Policy Council” to assist regulators and companies roll out AI.
The Australian Institute of Machine Learning says: “Australia urgently needs a formal, national strategy for artificial intelligence to ensure that we are net beneficiaries and not simply powerless recipients of this exciting and disruptive new technology.”