Last week the billionaire owner of an AI firm predicted that AI technologies could erase half of all entry-level office jobs. While Dario Amodei’s claim caused a stir, a respected CNN business analyst concluded it was a publicity stunt.
“Amodei is a salesman,” wrote Allison Morrow, “and it’s in his interest to make his product appear inevitable and so powerful it’s scary. [News website] Axios framed Amodei’s prediction as a ‘white-collar bloodbath.’”
These days we are used to seeing AI welcomed as some sort of magic tonic for the economy. Former Minister of Finance Steven Joyce, who (as I write this) is poised to become board chair of NZME (the publisher of the NZ Herald), repeats a familiar mantra in his Herald column, AI is key to unlocking New Zealand’s economic potential.
“The Global Financial Crisis and the Canterbury earthquakes bred huge uncertainty, and now people think they were quite small,” he opined. “AI is definitely big… This innovation is bringing huge expansions in the ability of all of us to do things that would have previously taken an army of software engineers.”
While I do not know anyone who thinks the effects of the Canterbury earthquakes were ‘quite small’, yes, AI is ‘definitely big’.
It has already transformed NZME’s masthead title. As former Herald editor Tim Murphy told RNZ: “AI – the robots, the machines – have taken over most of The New Zealand Herald website. They have recently put off about 15 online editors, people who would assess and process and curate on that website.”
Although itself hit with AI-led job losses, the media has been slow to write about the impact of AI on jobs. This weekend, The Guardian ran a long overdue feature headlined One day I overheard my boss saying: just put it in ChatGPT’: the workers who lost their jobs to AI.
Meanwhile, boosters like Joyce frame doubts over AI as a fear of change. If we don’t embrace ‘the excitement’ of AI, imply AI’s fans, we are out-of-touch losers, mojo-less techno-fossils.
But it is important to be aware of AI’s social, environmental and economic costs and flaws, along with its incredible abilities. Large Language Models (LLMs) like ChatGPT are plausible, not accurate. Their mistakes are not only plentiful but often difficult to spot – a big problem for organisations which depend on trust, like the media, or public health.
The Trump administration has announced huge cuts to medical research, claiming to have shifted focus to chronic disease. Their justifications are spelled out in the keenly awaited report Make America Healthy Again, launched last week. The only thing is, journalism students have discovered it relied on studies which don’t exist, making it almost certain the author of the report relied on LLMs over scientific expertise. “If a college student turned in this report as a paper, imagine the grade they would get for passing off such sloppy fabrications,” rued the excellent, human, New York Times science journalist Carl Zimmer. “This …after thousands of seasoned government workers, including scientists with lots of experience on things like chronic disease, were fired.”
Okay, it makes mistakes and kills jobs. But surely generative AI will have a big effect on productivity?
When I began my working life in the media, journalists hammered out stories on clattering typewriters and ‘artists’ like me used hot wax to stick ‘galleys’ together for the printer. Text corrections involved sticking cut-out apostrophes into scrolls of text. I can’t imagine doing that now.
In the 1980s and 90s, as labour economist Aaron Sojourner told CNN, “Computer adoption gave the world all kinds of tools that reshaped the labour market. But labour productivity grew just two to three percent.”
The IT ‘Productivity Paradox’ even has its own Wikipedia page. It’s strange, but true – as information technology soared in the 1970s and 80s, productivity growth slowed.
One of my oldest friends is a brilliant graphic designer. He has taken to sending me YouTube AI demos, with the horrified fascination of a man watching the demolition of his livelihood. In his last email, a YouTube video showed snippets of what looked like big budget movie scenes and immaculate television ads. They showed people singing “We can talk!” Except these people had never been born. They only looked human.
Film doesn’t just employ actors, but carpenters, truck drivers, stunt performers, caterers and experts in cinematography, lighting, art, special effects, makeup and wardrobe. Generative AI suddenly makes these jobs look like an indulgence.
It has been called an ‘oligarchic project’ sucking profit from workers to billionaires, an energy-draining, planet-trashing ecological disaster. It isn’t only artists who call it ‘the plagiarism machine’, or journalists who call it ‘the BS machine’.
A lawyer recently posted desperately on Bluesky: Tell your friends and family to stop using ChatGPT for legal advice. It’s happening constantly now. I’m always explaining to people why the advice this silly thing gave them is not real legal advice.”
One day soon, it might be.
And how will the economy grow, when the people replaced by AI can’t afford to buy anything?
• Jenny Nicholls
© Waiheke Gulf News Ltd 2025
