What exactly is AI doing to the labour market?

Why are some workers so worked up about AI while others aren’t?

Goldman Sachs estimates AI infrastructure spending could reach USD3 trillion to USD4 trillion by 2030. Nvidia offered guidance of USD65 billion revenue for next quarter, suggesting another 14% quarter-on-quarter step-up. Wall Street clearly thinks this could be a very large, long boom, and Nvidia wants us to think it’s a historical megacycle.
James Hennessy, Capital Brief, 20/11/25

A line published in last week’s post – “The outcome is less a wave of innovation than a gradual reduction in middle-tier administrative labour” – has haunting me ever since I wrote it.

Well, to be exact, I prompted it rather than wrote it. Which in appropriately tragicomic fashion, rather sums up the situation of we middle-tier administrative labour types.

Two-tier automation and its discontents
One reason there’s such furious debate about the impact AI is or isn’t having is that it’s affected some occupations a lot more than others.

If you’re, say, a builder married to a nurse who mainly socialises with tradesmen, and maybe the odd firefighter, it would be reasonable to wonder what all the fuss is about.

Conversely, if you’re, say, a content creator who’s married to an insurance claims processor who mainly socialises with market researchers, and maybe the odd academic, you could well be wondering why the prospect of mass unemployment isn’t being taken far more seriously by the powers-that-be.

Six labour market trends
With the third anniversary of ChatGPT’s launch imminent, I asked a range of AI platforms to draw on reputable sources and predict the most significant impacts AI will have on the labour market in 2026.

Here’s what it told me.

1. Career ladders will keep collapsing

Trend: Employers are formalising a shift that has long been underway: skills now matter more than degrees.

LinkedIn’s global analysis finds that skills-based hiring expands the talent pool roughly sixfold. A broader review of millions of postings shows the degree premium shrinking in fast-moving fields, even as credential inflation rises elsewhere.

Junior roles are fading as routine analytical and administrative work is absorbed by mid-career staff using AI tools. Upward mobility is becoming sideways mobility. Older workers are facing tighter constraints: job mobility drops sharply after 55, according to the OECD.

Impact of trend: Experience is becoming a scarce resource. Younger workers struggle to gain it; older workers struggle to repurpose it. Employers must now actively manage internal mobility rather than assume it will occur automatically. Governments are coming under pressure to rebuild early-career pathways the education system once supplied.

2. There will be plenty of jobs going in the care economy

Trend: The steep decline in the ratio of working-age people to older dependants continues. Australia alone needs at least 110,000 more aged-care workers within a decade.

Immigration, which has provided a solution to the issue for decades, is now tightening everywhere. Burnout and attrition in the care economy remain high. While some displaced white-collar workers will transition into care roles, uptake will stay limited unless pay, training and career structures improve substantially.

Impact of trend: Expect governments to expand subsidies, shorten qualifications and create targeted visa channels for care roles. Care shortages are likely to become a growing economic and political issue throughout the second half of the 2020s.

3. Productivity gaps will widen

Trend: IMF analysis shows productivity growth is clustering around a minority of firms with the capital, digital infrastructure and capacity to deploy AI at scale. OECD research shows the same pattern with ‘frontier firms’ growing nearly nine times faster than laggards.

Two workers with the same job title can now inhabit entirely different labour markets depending on their employer’s ability to adopt AI effectively.

Impact of trend: Inequality is shifting from being about occupation to being about employer. High-productivity firms attract scarce talent and set wage expectations; low-productivity firms struggle to retain staff or offer mobility. Policymakers face a two-speed economy that traditional labour levers cannot easily influence.

4. Things won’t improve for young people

Trend: The OECD reports that around 14% of 18–24-year-olds across member countries are NEET – neither in work nor education. The issue is not simply youth unemployment but also youth under-placement. That is, graduates working below skill level because entry-level tasks, the work that used to build capability, have been automated.

As Peter Turchin’s ‘surplus elites’ theory warns, this is ominous. Societies that produce more credentialled young people than they can absorb tend to face heightened political volatility.

Impact of trend: Young adults will struggle to achieve independence. Employers will weaken their future talent pipeline. Governments will come under pressure to rebuild early-career experiences and regulate AI substitution in junior roles. Politically, frustration among downwardly mobile graduates will become harder to ignore and contain.

5. (No longer abundant) labour will become a geopolitical asset

Trend: Supply chains are being reorganised around demographic capacity and political alignment, not cost. Countries with large, young workforces — Mexico, Vietnam, India, Poland — are moving up in the world. Ageing economies – China, Japan, South Korea, Europe – are looking to accelerate automation to avoid slipping down the food chain.

Impact of trend: Jobs flow toward jurisdictions with demographic resilience and predictable governance. Migration becomes narrower and more skills-specific, especially in engineering, advanced manufacturing and care. Labour is no longer just filling market gaps, it’s being directed by national strategy to build long-term strength and resilience.

6. Full-time employment stops being the default

Trend: Across advanced economies, full-time roles are slowly eroding. Firms increasingly rely on fractional specialists, multi-role portfolios, algorithmic scheduling and internal project marketplaces. AI enables more granular task allocation and reduces dependence on fixed roles.

Older workers will semi-retire into advisory positions; high-skill workers will become ‘micro-firms’ operating across projects; lower-income workers will face more volatile, demand-driven hours.

Impact of trend: Income stability will weaken. Access to benefits tied to full-time work will become uneven. Employers will gain flexibility but lose continuity and institutional knowledge. Governments will need to re-engineer tax and welfare systems built on the assumption of stable full-time jobs.

In conclusion
I used ChatGPTPlus for the final drafting (prompting?) of this piece. I only asked it to synthesise the output of its now-near-omniscient digital siblings (the free versions of Perplexity, Claude, DeepSeek, Grok and Gemini) and spit out six trends. Nonetheless, it anticipated my needs and also provided a conclusion.

A long-winded and flowery conclusion.

Now, I could have just bashed out a ‘Halve the length and make it less long-winded and flowery’ prompt and left you with the results, dear reader.

But despite the hour being late and the year long, I decided to kick it old-school wordsmith style.

I give you the results of me spending five artisanal minutes halving the length of the AI-generated conclusion. As well as making it – by this blog’s admittedly loose standards – less prolix.

What This Means for You
You are no longer on a predictable escalator. You’re navigating a grid. Progression depends on where you work, not just what you do.

None of this fits the beliefs workers were raised on. The assumptions that qualification leads to opportunity, experience leads to security, and effort leads to a stable future. That was the social contract of the second half of the 20th century. That social contract is beginning to look worryingly frayed. The labour market still functions. It just no longer functions for the life model people expect.

Governments will adjust, firms will adapt, and new pathways will appear.

But the adjustment costs fall unevenly. The changes mostly won’t feel like sudden shocks. They’ll feel like drift: ladders shortening, titles fading, hours shifting, expectations changing.

The people who adapt fastest are the ones who stop waiting for the old deal to return.