It feels as if AI is everywhere – whether it’s those intrusive bots on every website or every headline about how it’s either going to be a boon for humanity, or end us.
But is AI everything its evangelists and doomsayers say it is? Perhaps not.
Despite the hysteria, in 2024, the respected computer scientist and commentator John Naughton predicted that AI was a bubble waiting to burst. I’d agree.
There’s every indication it’s much like the dot com bubble, with significant economic, environmental, political, and geographic risks that may burst the whole mirage before we even get to liberation or annihilation.
The incredible sums being thrown around – like OpenAI seeking $7 trillion dollars in investment (around a fifth of US GDP, and three times Australia’s) and both OpenAI and Anthropic being valued ahead of their Initial Public Offerings (IPO) at nearly a trillion dollars each – are vastly disproportionate to their ROI (return on onvestment): both companies are running colossal losses, with little indication when or how they’ll start turning profits.
Serious governance issues aside – especially in regard to Elon Musk’s much-hyped USD 1.78 trillion IPO, which has raised questions from many investors – these hyperinflated valuations have distorted the US stock market, with just ten AI stocks constituting up to 45 per cent of the S&P 500, with many pension and index-linked funds exposed.
While markets are at historical highs, anaemic underlying global economic growth and rising unemployment suggests a likely global recession. If any of those stocks fall or those very interconnected corporations fail, forget America sneezing and the world catching a cold: it will be global economic Ebola.
While record numbers of white-collar workers have been laid off because of AI’s supposed greater productivity, a recent US survey found that many workers are finding that the AI’s often error-riddled, hallucinatory ‘workslop’ requires much more work to correct, resulting in significant productivity losses.
And tech layoffs of programmers aren’t entirely due to AI, but because tech companies over-hired during the pandemic when more of us had to be online – they started in 2022, before ChatGPT was released. Zoom’s share price has fallen 80 per cent since its lockdown peak.
But even if AI did result in widespread white-collar and creative sector job losses, there are many jobs it can’t replace, especially in human-centred roles, with many young workers, increasingly suspicious of and resistant to AI, now taking up trades or working in healthcare.
And while a landmark 2015 Deloitte study of 140 years from the first Industrial Revolution found that technology created more jobs than it destroyed, replacing legacy occupations with newer, better paid ones (such as tenant farmers becoming loom machinists, or actuaries becoming data scientists), if AI did cause the predicted 40 per cent unemployment, what government would tolerate – much less survive – the subsequent social unrest and economic devastation?
Many communities are pushing back against AI’s environmental damage, with data centres guzzling massive amounts of electricity and water, overloading local infrastructure and potentially surging power and water rates – especially galling when most big tech companies don’t pay tax in most jurisdictions.
But that’s not the most critical geographical issue. Over 90 per cent of the world’s most advanced semiconductors required for AI are made by one company, the Taiwan Semiconductor Manufacturing Company.
And China, with claims on the island, holds half the world’s rare earth reserves required for modern chip manufacture, mining about 70 per cent of the global supply, and accounting for 90 per cent of all processing and refining.
Donald Trump’s clumsy tariff war against China – and misjudged war-not-war against Iran – have badly exposed the US economically and militarily.
Just as the Islamic Republic has used the Strait of Hormuz as leverage, so the People’s Republic has used its rare earth near-monopoly to extract tariff and other concessions.
With Trump wasting nearly half of America’s missile stockpile on his Iran ‘excursion’, predictions it will take a decade or more to replenish them, and with most of the US Navy blockading the Persian Gulf, Taiwan is ripe for the taking.
Although having seen the disastrous invasions of Ukraine and Iran, the challenges of an island invasion, and the People’s Liberation Army not seeing combat since the brief 1979 Sino-Vietnamese war, it’s unlikely Xi Xinping will risk it.
But still, it’s as significant a potential threat to world security as the Hormuz blockade has become for energy security.
Already, markets have started cooling on tech stocks over concerns about how they’ll keep their big promises and fund their exorbitantly expensive expansion plans.
So there are signs that long before we get to a Terminator-style Skynet extinction, the AI bubble will pop – although not before its snake oil salesmen like Open AI’s Sam Altman and his frenemy Elon Musk have taken their very juicy cuts.
But now’s the time to put in place the regulatory and legal safeguards we neglected to enact before the toxic bin fire of social media, to ensure neither worst case scenario happens.
And with the Albanese government prodded into action this week, by independent Senator David Pocock, to regulate data centre energy usage and tax them the way gas and tech companies haven’t been so far, hopefully that will happen sooner rather than later.
Sunil Badami is a writer, academic and broadcaster. He’s appeared in nearly every major Australian media outlet, and is a regular on ABC Radio and TV.
For four decades The Echo has printed the stories some people loved, some people hated, and some pretended not to read. If you want us to keep telling the truth, the real truth, not the sugar-coated version. We’ll need your support to keep the presses rolling. 
