If you have been waiting for the AI mass-layoff wave the internet keeps promising, a Harvard Gazette analysis published on 29 September makes an awkward observation: it hasn’t arrived. Unemployment sits around 4.1 percent, buoyed by healthcare hiring, and for every article announcing AI-attributed job cuts there is another calling the attribution “AI-washing” — companies borrowing the technology’s vocabulary to dress up cuts they would have made anyway.
The economists behind the piece think the calm is real but temporary — and their reasoning matters more for careers than the doom headlines do.
Today’s numbers say little about the next decade
Kennedy School economist Doug Elmendorf, who co-authored a National Bureau of Economic Research working paper this year with Karen Dynan and Brookings’ Louise Sheiner, puts it bluntly: “What we’ve seen so far in the labor market from artificial intelligence has very little predictive power for what we’re going to see in the labor market because of AI in five years, or 10 years, or 15 years.”
He expects “significant job losses” over a couple of decades but believes most affected workers will find new work. The NBER paper’s modelling scenario puts about 3 million people — roughly 2 percent of the US labour force — out of work at any given time in one of its faster-automation scenarios, compared with the 1.5 to 2 million total jobs the researchers cite from the 2000s “China shock.” Their recommended preparations read like insurance: wage insurance for displaced workers, subsidised employment, shorter work weeks, expanded retraining.
Why firms aren’t cutting — the implementation gap
Harvard Business School’s Joseph Fuller, working with Accenture Research, built an AI model that analysed work tasks across all job categories and found 41 percent of all work tasks can today be automated or augmented by AI. Yet adoption lags the technology. Only about a third of corporate AI experiments succeed, he says — often because of bad data or untrained staff.
“Companies are not training people formally, so employees are basically using it like glorified Google, and they’re not getting much productivity out of it,” Fuller told the Gazette. That clumsiness is currently a jobs shield. He expects it to erode as cheaper, more specialised AI agents arrive.
Colleague Raffaella Sadun, who runs HBS’s Digital Reskilling Lab, argues the bigger risk is executives treating human capital as optional: automating people away trades short-term savings for the loss of firm-specific knowledge, what she calls “a tremendous trap.” At her lab, retraining for AI-augmented roles is tested the way a firm would test any capital investment — because it is one.
The constructive idea: redesign the first rung
The most career-specific idea in the piece belongs to Fuller. He works with executives who hear that a fresh computer-science graduate offers less value than before — but who also know there is no senior software engineer without first a junior one. His fix is to change what entry-level jobs contain rather than to stop hiring juniors.
At one large tech company, he studied what senior engineers’ performance reviews said they lacked: sales and product management experience. “So rather than stopping hiring junior software engineers, we changed the job description so those hires are still doing some software engineering, but they’re also getting regular structured introductions to sales and product management,” he said. His shorthand: the entry-level job may now look the way a second or third job did five years ago.
That reframe — entry-level hiring responding to AI in both directions — fits the wider data. A Ramp and Revelio Labs study of 21,000 US firms found the biggest AI spenders grew entry-level headcount 12 percent, while McKinsey’s new mobility report projects most US workers will need to build new skills even if they keep their occupations. In New Zealand, the same squeeze is visible from the other end: graduates report hundreds of applications per role as AI-written applications flood recruiters, and the Reserve Bank has warned AI could soften job growth for young workers in the short term.
The honest summary
Nobody in the Harvard piece declares the danger over. Elmendorf’s line is the opposite — prepare for changes bigger than anything seen so far, because you would rather not be surprised. Fuller’s and Sadun’s contribution is to show what preparation looks like inside a single company: it is not a layoff memo and it is not a prayer. It is a rewritten job description that keeps the first rung of the ladder in place while teaching the things AI will not store for you.
For anyone early in a career, the version that survives contact with this evidence is unspectacular: learn the tool, but more importantly put yourself where structured work and judgment meet — because that is what the redesigned rung is being built to teach.
FAQ
Have AI mass layoffs happened yet? Widespread AI-attributed layoffs have been announced — more this year than last — but economists including Harvard’s Doug Elmendorf say the aggregate US labour market data does not yet show mass displacement, and some announced “AI” cuts are contested as AI-washing.
Will AI take entry-level jobs? Tasks that are codified and checkable are most exposed, which is where entry-level work traditionally lives. But firms that adopt AI heavily show mixed hiring outcomes — a Ramp/Revelio Labs study found the biggest AI spenders grew entry-level headcount 12 percent.
What is “redesigning the entry-level job”? Harvard’s Joseph Fuller describes restructuring junior roles so new hires still learn core skills while getting structured exposure to adjacent experience — sales, product management — that future senior roles will demand.
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
