Perils and promise of AI’s brave new world
Labor is banking on a productivity surge from these disruptive tools, but Australians are among the most apprehensive in the world about the technology.
On his way to the G7 summit in Canada a fortnight ago, Anthony Albanese had a layover in Seattle to attend an investment event at The Spheres on the Amazon campus.
It might have been “a nice sunny day”, as the Prime Minister’s host put it, but “the cloud” was omnipresent. Amazon Web Services announced it was investing $20bn across five years in Australia to support artificial intelligence and cloud computing for customers, including the Commonwealth Bank, while claiming it would pave the way for start-ups to become the next Atlassian or Canva.
AWS chief executive Matt Garman declared it “the largest investment ever announced by a global technology provider in Australia”, while Albanese said the two data centres (and three new solar farms) would allow local players “to take advantage of the revolutionary opportunities” provided by AI.
Generative AI is the zeitgeist, bringing together civilisation’s vast store of data with unprecedented computing power. In response to prompts entered by a human into a computer program known as a chatbot, this predictive tool can analyse huge datasets (basically, the entire internet), finding patterns and filling gaps, to create text, images, audio, video or data,
Even central bankers can’t contain their excitement. “The economic potential of AI has set off a gold rush across the economy,” the Bank for International Settlements said a year ago, noting the “breathtaking speed” of adoption
The November 2022 release of OpenAI’s ChatGPT and its widespread adoption was a game changer. It’s now the world’s most popular chatbot with an estimated 300 million active weekly users. OpenAI’s chief executive Sam Altman says his company and its rivals “are building a brain for the world”.
A year after its arrival, more than one-third of US households had used ChatGPT. To reach that concentration it took smartphones four years, social media five years, the internet seven years and electric power and computers 13 years.
What these “stochastic parrots”, based on large language models, do well is write computer code and memos; the essays are OK by the standards of dim undergrads but they’ll never come close to creating the ecstasies of Shakespeare, Donne, Dylan or Cave.
As companies train workers in AI through “boot camps” (as we have at this newspaper) there’s also passive adoption and integration (via updates of third-party software). This column dutifully consults Dr Google; rather than simply searching the internet as asked, the engine acts like a tenured professor, slipping in a mini-lecture before revealing the results requested.
Naturally, given Silicon Valley’s modes, its unbridled boosterism and bottomless pockets of the plutocrats in an ever-expanding multiverse, the hype around the next AI iteration (machines with full human-like cognitive capabilities) is immense, like a Donald Trump brag to the power of a billion.
The flip side is normal people are unsettled by these all-conquering algorithms that learn as they go, invading privacy and gobbling up data, energy and water, as well as entry-level jobs, as they infiltrate every area of life from finance to medicine, art to relationships.
AI tools have been created by cancer researchers, co-led by the Garvan Institute of Medical Research, to detect biological patterns in cells within tumours. As eSafety Commissioner Julie Inman Grant warned this week, as well as promise, the evolving and relatively cheap technology creates peril by enabling child sexual exploitation material online and captivating our children with AI companions.
But Australians have been reluctant to embrace AI because of mistrust of Big Tech, the speed of its uptake, their cavalier attitude to copyright and creatives, and fears about job losses. In January, EY’s AI sentiment report found Australians among the most apprehensive in the world about the technology.
Our companies are behind the play. According to Committee for Economic Development of Australia chief economist Cassandra Winzar, we rank a poor 54th (out of 69 nations) on companies’ use of digital tools and technologies in the latest global competitiveness report by the Swiss-based Institute for Management Development.
Winzar says Australian firms could be left behind in the AI rush. “Our companies are risk-averse, slow on the uptake of new technologies and slow to adopt dynamic market capabilities,” she says. “We often quickly identify the need to adopt but we’re not willing to put ourselves on the line, make the changes and reap the advantages.”
She says there’s a lack of tech expertise on boards, which are over-indexed with lawyers and accountants. As well, there’s little slack in local firms, which inhibits strategy and implementation, while a fall-off in dedicated training risks leaving workers exposed.
As Labor tells it, generative AI is one of the most promising enablers for growth, jobs and productivity. Minister after minister is urging employers and workers to “lean into the opportunity”. Techno optimists in the academy say AI is not merely a tool, it’s an entire system.
New OECD research is cautiously optimistic about whether AI is a “general purpose technology”, like electricity or the internet, that will lead to widespread benefits. The Paris-based think tank’s review notes that AI appears to exhibit the defining characteristics of GPTs, namely pervasiveness, continuous improvement over time and innovation spawning.
“While productivity gains may not materialise immediately, the evolution of earlier GPTs seems to provide encouraging signs that generative AI could lead to substantial improvements in productivity in the future,” it says.
Nobel laureate Paul Krugman is sceptical. “We’re not yet seeing the productivity surge,” the US economist told Martin Wolf of the Financial Times in a lively exchange about AI hype and realities. It took 40 years, Krugman says, for businesses to figure out what to do with electricity. And then it was transformative: production changed, as well as jobs, land use and cities.
The Productivity Commission argues AI adoption involves both augmenting and automating work tasks, which increases labour productivity and frees up workers’ time. A 2024 study by the McKinsey Global Institute estimated up to 62 per cent of Australians’ work time could be automated, although this varies by occupation.
“AI can substitute for workers’ specific tasks, potentially improving the quality of work for employees,” the PC told a Senate committee. “But more typically, AI is expected to enable more efficient use of the existing workforce, particularly in areas where there are skill and labour gaps.”
An optimistic note was struck by the International Monetary Fund in its April exploration of healthy ageing among baby boomers. Creators, analysts and decision-makers are likeliest to thrive and survive in the new era, as long as there are lifelong skilling programs, because of “the complementarity of their skills with AI”. “Unskilled workers may struggle to keep their jobs or manage successful job transitions,” the IMF said.
This week Productivity Assistant Minister Andrew Leigh presented evidence that job growth was most rapid among firms that were early adopters of AI. “This means that the biggest employment risk from AI may not be job displacement – it may be working for a business that doesn’t adopt it,” he wrote in an email.
Jim Chalmers has asked the PC to conduct five inquiries into the pillars of prosperity, one of which is data and digital technology, including enabling AI’s potential. The interim report is due ahead of the Treasurer’s roundtable in August.
Submissions to the PC cover the gamut of tech lobby evangelism about 200,000 new roles by fully embedding AI into end-to-end processes; dire warnings from creatives about the erosion of copyright protections; and worries about AI’s overuse from our oldest university (leading to “cognitive atrophy”) and engineers fearing about the competency of recent graduates.
Chalmers told the National Press Club last week the government wants to capitalise on the huge gains from AI, “not just set guardrails”. “We want to get the best out of new technology and investment in data infrastructure in ways that leverage our strengths, work for our people and best manage impacts on our energy system and natural environment,” he said.
The AI rollout has caught regulators’ attention. This year the mega platforms will spend about $400bn on generative AI. It may be years before they reap big returns from these products, “raising questions about what sources of revenue will be used to eventually recoup these costs”, the Australian Competition & Consumer Commission said in the final report of its five-year digital platforms inquiry.
Leigh argues regulation should follow a principles-based approach. “Start by applying existing laws,” he told the McKell Institute this week. “Where those fall short, make technologically neutral amendments. Only if these approaches are insufficient should AI-specific rules be considered. The goal is to protect the public while allowing productivity-boosting AI innovation to flourish.”
Labor has displayed an abundance of caution in formalising new laws. Some argue the technology is not new and current laws may be enough. It won’t be easy to find a sweet spot between a sceptical public and tech’s libertarian tendencies. Or to dispel the hype.
The next frontiers are artificial general intelligence or machines with full human-like cognitive capabilities; Meta founder Mark Zuckerberg and others are pursuing superintelligence, which is a few levels above Elon Musk, before we reach what Altman calls a “gentle singularity” of an intelligence explosion. “Humanity is close to building digital superintelligence,” Altman wrote on his blog this month, while claiming “in some big sense, ChatGPT is already more powerful than any human who has ever lived”. So is a pocket calculator when it comes to maths.
Can this pumped-up autocorrect fix a leaky pipe, tag Nick Daicos at the MCG or take out the garbage on Sunday night? The bot told Inquirer it “currently lacks the physical capabilities required” to perform these tasks and besides “these activities necessitate human intervention or specialised machinery”. It’s working on it.
More Coverage
Add your comment to this story
The pressure on defence spending will not abate
The debate should not be about Donald Trump or NATO, but the growing challenges in this region.
PM and Wong: ferals who pretend to be moderates
At every turn in foreign policy their radical anti-establishment ideology is exposed.
To join the conversation, please log in. Don't have an account? Register
Join the conversation, you are commenting as Logout