We’ve reached the end of the beginning of the AI transformation

The AI rubber is starting to hit the road

Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning.
Winston Churchill, 10/11/42

In a rare and regrettable lapse into self-doubt, I briefly considered the possibility I was wrong about AI in a post at the end of last year.

The tl;dr is that I’d spent all of 2025 attempting to convince readers AI was going to rapidly and profoundly disrupt labour markets, economies and societies. (Societies already riven with widening class, ethnic, religious, generational and gender conflicts.)

But six months ago, looking at the scoreboard – unemployment and underemployment rates, job ads, wage growth, productivity improvements, media and public attention devoted to AI – I had to begrudgingly concede that it was difficult to find solid evidence AI was rapidly and profoundly disrupting anything.

As I wrote at the time:

If AI were as transformative as its boosters insist it is, it would inevitably come to dominate the news and political agenda. But while AI remains a hot topic, it’s not even as hot as, say, housing, immigration or cost of living pressures… maybe I’m wrong and the dwindling band cynics still insisting AI is an overhyped nothingburger will ultimately be proven right. I imagine things will be much clearer at the conclusion of 2026, so I’ll do a follow-up post then.

Well, it’s mid-2026 and the evidence that AI is in the process of rapidly and profoundly transforming labour markets/economies/societies is becoming ever more suggestive.

All AI, all the time
If AI was changing the world – if it was like the Industrial Revolution, except at 10X the speed and with 10X the impact – you’d expect AI to be highly salient.

You’d expect the media to devote ever-increasing amounts of attention to it. For it to be what we Australians call a ‘BBQ-stopper’ – an issue that comes up spontaneously when people gather to socialise. You’d expect some individuals to decide they need to go all Sarah Connor to prevent or delay an imminent AIpocalypse. You’d expect commencement speakers to be booed when they mention AI. You’d expect “fiscally conservative, socially liberal” professional-managerial class types to develop a sudden interest in joining a union. You’d expect employees having their roles automated away to lash out at their employers. You’d expect people to be seriously worried about their future career prospects, or at least those of their kids. You’d expect captains of industry to be making regular pronouncements about the likely course of the AI transformation. Hell, you might even expect the Pope to churn out an encyclical that frames AI as the defining social question of the age and warns it can be dangerous when driven by profit, surveillance and concentrated corporate power.

In short, you’d expect it to be all AI, all the time.

And now, it pretty much is.

An elite-human-capital perspective
Elite human capital is usually too busy undertaking elite, well-remunerated activities to pen reflective essays for publications such as The New York Times.

But final-year Stanford University student Theo Baker recently addressed the topic of What A.I. Did to My College Class.

Here’s Baker on the salience of AI for students attending top-tier universities:

A.I. is everything. We talk about it at the dining halls and in history classes, on dates and while smoking with friends, at the gym and in communal dorm bathrooms.

Here’s Baker on how AI has now transformed the higher education experience:

But for many who came to Stanford — just four years ago! — when a degree seemed like a guaranteed ticket to a high-paying job, the door has been slammed shut. For all of us, A.I. has permanently changed how we think and behave.

Here’s Baker foreshadowing how his generation of hyper-ambitious geniuses might seek to monetise AI:

Stanford already had a shaky reputation for integrity when I arrived in 2022. It was the origin place of the Theranos fraudster Elizabeth Holmes (now serving a 10-year prison sentence), the crypto fraudster Do Kwon (now serving a 15-year prison sentence) and the founders of Juul (which was forced to pay billions for getting kids hooked on vapes)…

Many of my classmates arrived idealistic and hopeful, but among the strivers seeking a path to fortune, hustle culture was the accepted way of life. Now A.I. has made deception easier and more remunerative than ever before. Cheating has become omnipresent. I don’t know a single person who hasn’t used A.I. to get through some assignment in college.

Here’s Baker on the downsides of outsourcing your thinking, as we are all increasingly doing:

Relying on A.I. for cognitive tasks can reduce one’s own intellectual capacity and resilience… Sure, a robot can lift 600 pounds much more easily than I can — but that doesn’t much help me if I’m trying to work out. The same goes for the thinking exercise of education.

Here’s Baker on Gen Z becoming the ‘AI generation’ in the same way Gen Y were the ‘smartphone generation’:

It’s becoming baked in, shaping our generational character. We are a digital generation, growing only more attached to the virtual world… The internet has already allowed us to feel more connected than ever while becoming lonelier than ever. A.I. lets us cut out the human part of human interaction entirely.

What to train for, where to work, what to do?
A prominent Australian academic also wrote a column this week questioning the value of a university degree in the AI age. In a piece titled I’m an academic, but I’ve told my stepdaughter to think twice about going to university, Dr Kylie Moore-Gilbert says the quiet part out loud:

In the three years since ChatGPT was released we have arrived at a point in which all of Australia’s universities are committing widespread, industrial-scale fraud. The students who began their studies back then are now graduating and entering the workforce, and we’ll soon begin to see the results of a real-time experiment in degree by GPT.

Dr Moore-Gilbert is worried by the likely results of that experiment.

Let’s grossly underestimate the numbers of students who have GPT degrees and put the figure at a measly 10 per cent (the reality is probably well over 90 per cent).

Should we accept that even 10 per cent of the future engineers of our roads and bridges didn’t earn their qualifications... Maybe the fraud doesn’t matter if these folks can continue outsourcing their cognitive functioning after they enter the workforce… [but] that sounds to me like an easy argument for the replacement of those workers with the same AI agents who sat their degrees for them. Integrating AI into the foundations of human education will not only lead to mass brain rot; it will likely hasten the AI jobs apocalypse.

Ah, the old AI job apocalypse I’m always banging on about.

When I churned out the aforementioned December post, there wasn’t yet convincing evidence that AI was increasing unemployment or underemployment. The aggregate labour market data hasn’t yet moved decisively, but targeted studies are starting to find employment effects in high-exposure occupations. No less an authority than Goldman Sachs expects 2026 to be the year the broader numbers shift.

We may still be some months from having conclusive proof AI is significantly impacting the labour market, but the evidence we currently possess is increasingly persuasive.

Which brings me to some other Antipodean news stories.

AC/DC guitar tech turned logistics software gazillionaire Richard White was in the news this week because one or more of the 2000 employees he’s replacing with AI made violent threats against the CEO overseeing the culling of a third of his company’s workforce.

That White thinks nothing of making observations such as, “It doesn’t take much effort to convince people, in the end, that they’re stupid to be paying $100 for labour when you can pay $2 for the AI,” probably hasn’t improved staff-management relations at WiseTech.

Matt Comyn, the CEO of Australia’s biggest bank, published an op-ed in the nation’s business paper making essentially the same point as White, albeit in more diplomatic language.

The original official narrative around AI was that it would only ever augment humans and never be able to replace them. When it soon became apparent that was nonsensical, the narrative was tweaked to, “You won’t be replaced with AI but you might be replaced by someone using AI.” We now seem to have moved to a grimmer if more honest narrative – “You will soon be replaced by AI and it would be irresponsible of us to pretend otherwise.”

Comyn’s piece is titled Pretending AI doesn’t mean change won’t protect workers. In it, he observes:

AI will have workforce consequences throughout the economy, and they should be faced directly. No one knows exactly how work will change over the next three years, let alone the next decade. Some tasks will be automated, some roles will be reduced in number and others will grow...

This will create opportunities for many people, but it will be demanding for everyone. Pretending otherwise does not protect workers. It only ensures they are surprised later…

When a role disappears, it affects a household budget, a mortgage, a career plan and, for many people, a sense of identity. Saying AI is good for national productivity does not make loss painless. But pretending every role can be preserved would not be fair either. An employer of our scale has a responsibility to avoid false reassurance and give people the best possible chance to adapt.

The End of the Beginning
The 3.5 years since ChatGPT dropped have had a ‘Phoney War’ vibe. The Phoney War was the first six months of WW II when Britain and France were officially at war with Germany, but conflict was minimal. There were propaganda-leaflet drops, a few merchant ships sunk by U-boats and the occasional skirmish. But especially in comparison to what was to come, nothing much happened.

During the Phoney War stage of the Great AI Transformation, there have been three schools of thought.

The techno-optimists conceded AI would create profound disruption but insisted the positives would far outweigh the negatives. In particular, techno-optimists have loudly insisted fears of mass unemployment are misplaced.

Like the techno-optimists, the techno-pessimists believe AI will profoundly transform economies and societies. But given the lessons of half a century of neoliberalism, the pessimists fear elites will harness AI to redistribute even more wealth and power upwards.

The AI sceptics argue that AI is an overhyped nothingburger.

I’ll leave it to you, dear reader, to decide whether the optimists, pessimists or sceptics have been proven most prescient as the first phase of the AI transformation draws to a close.

I’ll wrap up by conceding that the AI transformation is yet to produce a labour-market Dunkirk, but asserting the Phoney War phase of the Singularity is over.