What if there’s a Rough Singularity?

When a labour market changes, so does the economy and society it’s attached to

Photo by Steve Johnson on Unsplash

Power concedes nothing without a demand. It never did and it never will.
Frederick Douglass, 1857

As we roll out more generative AI and agents, it should change the way our work is done… It’s hard to know exactly where this nets out over time, but in the next few years, we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company… Those who embrace this change, become conversant in AI, help us build and improve our AI capabilities internally and deliver for customers, will be well-positioned to have high impact and help us reinvent the company.
Amazon CEO Andy Jassy, Business Insider, 18/6/25

What sort of society do we want to live in when AI hits its stride? That is the conversation we need to be having – thinking about the parts of our lives we want to automate and those we don’t. From the vantage point of public policy: we need to design the settings that both stimulate innovation where we most need it and ensure that the economic gains we stand to make from new technologies are broadly distributed through effective institutions of tax, a decent safety net and fair wages for those jobs for which humans remain irreplaceable.
Grattan Institute CEO Dr Aruna Sathanapally, SMH, 18/6/25

When ChatGPT first dropped, almost three years ago, everybody was quick to point out “it could never replace a human”. Soon enough, that kind of talk died down, but the conventional wisdom still held that AI would only ever “augment”, rather than replace, human workers.

Things took a darker turn once everyone started nodding sagely and proclaiming, “AI won’t take your job, but somebody using AI might!” Nonetheless, the implication was that anybody conscientious enough to do their AI-tool-learning homework could expect to remain employed. The ‘learn AI and everything will probably be OK’ narrative hasn’t entirely disappeared. There’s an echo of it in Jassy’s reassurance that change-embracers who become AI conversant will be “well-positioned to have a high impact and help us reinvent the company”.

But that narrative now seems to be being replaced by yet another one that goes something like this: “Yes, your prestigious and/or well-paid white-collar job is about to get vapourised. But don’t worry because it’s not like the powers-that-be would ever let anything bad happen to those no longer considered economically useful.”

Last week, riffing on the new official line, Sam Altman reassured increasingly nervous humans that they could look forward to a Gentle Singularity.

Rapid mass automation will happen – indeed, is already happening – Altman conceded. But he suggested the short-term pain (There will be very hard parts like whole classes of jobs going away) would be mitigated by better-funded safety nets (the world will be getting so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before).

Anyhoo, humanity shouldn’t worry about a slightly bumpy transitional period over the next few years because life on the other side of the benign technological tipping point will be full of marvels. Unimaginable scientific breakthroughs, off-the-charts abundance, ample leisure time, maybe even immortality:

The rate of new wonders being achieved will be immense. It’s hard to even imagine today what we will have discovered by 2035; maybe we will go from solving high-energy physics one year to beginning space colonization the next year; or from a major materials science breakthrough one year to true high-bandwidth brain-computer interfaces the next year.

Hmmmmm.

Causes for scepticism
Sharp-eyed readers will have observed that Altman, like Jassy, worded his screed in a slippery manner. For instance, he’s not saying displaced workers will be looked after, let alone calling for concrete actions to be taken ASAP to ensure this happens. Instead, he airily proclaims that so much wealth will be floating around in the highly automated future that new policy ideas will become feasible.

There’s no clarity around what these intriguing new policy ideas will be (Tax holidays for Silicon Valley’s God Emperors? Legally recognised polycule marriage? A UBI?) or even any confidence that they’ll be implemented. But the vibe of the thing is nobody should be too upset about being rendered technologically obsolescent in the nearish future because they can look forward to an unimaginably joyous sci-fi future.

Interestingly, the general public has been consistently more pessimistic about AI than techno-optimists in the tech industry. On my reading of the polling data, people believe the part of the story about their jobs potentially being automated away. But they remain unpersuaded about the happy ending, where AI’s benefits are distributed equitably.

Inexplicably, the citizens of nations that haven’t endured four decades-plus of increasing wealth inequality are rather more optimistic about the benefits of AI, shall we say, trickling down to the little guy.

A couple of proto-Singularities to consider
We may just have a gentle singularity. But if we do, that will be the historical exception rather than the rule.

Everybody eventually benefitted from the printing press. Nonetheless, it was the catalyst for Europe’s bloody wars of religion, which claimed the lives of millions.

Likewise, everybody ultimately benefitted from the industrial revolution. But it was a Dickensian nightmare for a large swathe of the population for at least a century, and arguably a century and a half.

I don’t believe Altman and Andreessen are misrepresenting the possible upsides of AI. But I’d argue the techno-optimists are glossing over how difficult the transitional period will be, even in a best-case scenario where everything goes well.

So let me encourage you, dear reader, to consider what a non-gentle Singularity might look like.

The Rough Singularity
There will be a lag between the start of mass automation and society's response. I’ve argued that the mass automation of ‘cognitive labour’ roles has already begun, and some demographics, such as content creators and grads, are already feeling the impact.

If we do see the mass automation of existing roles without the creation of new ones, prepare for the following.

The collapse of many marriages
It is a truth universally, if usually tacitly, acknowledged that women judge men on their resource-provision capacities. If a man can provide plenty of resources, he typically doesn’t lack for female companionship, no matter how physically unprepossessing and personally obnoxious he may be. Likewise, if a man cannot provide resources, he’s unlikely to get much female attention, no matter how many admirable qualities he may otherwise possess.

If you’re interested in why couples are much more likely to break up after the man – but not the woman – loses their job or business, you might enjoy this Time article bracingly entitled, “Men Without Full-Time Jobs Are 33% More Likely to Divorce”.

Deindustrialisation was strongly correlated with rising divorce rates and family breakdown. AI-driven mass automation of white-collar roles will probably play out similarly. When poverty comes in through the door, love flies out the window, as our German friends have long observed.

The collapse of the real estate market
As is often observed, especially by non-Baby Boomers, housing is expensive. Regardless of what transpires with AI-driven automation, people will still need a roof over their heads. But there is the small matter of where the money for the rent or mortgage payment is coming from.

And what happens to the banks and wider real estate industry when the defaults start? It’s not my area of expertise, but my fellow Substacker Will Rackham has helpfully mapped out how things will play out if the unemployment rate hits double digits and then keeps heading north:

Even at 15% [unemployment], we lose the ladder. At 30%, we lose the floor. Property becomes a liability.

It’s a domino effect. As buyers vanish, property stops behaving like an asset. With so few sales, there are no benchmarks to value homes. That’s when the state steps in with price controls, emergency liquidity, and eventually, AI-driven housing rationing - administered, no doubt, with typical levels of government surgical precision…

Housing becomes rationed like water. Nominal value evaporates. AI allocates based on priority: health workers, carers, energy engineers. A kind of bureaucratic housing socialism emerges, but one managed not by councils but by code.

The collapse of the university system
For the foreseeable future, people will presumably continue to send their children to primary school. Most likely, they will continue encouraging them to complete high school. But as I’m hardly the first to point out, bloated university systems across the first world would seem to be heading into a world of pain, not least because the universities had lost the general public long before ChatGPT materialised.

Ironically, the insufferably arrogant universities are most likely to survive the imminent culling, while all the second-tier ones that have popped up since WWII are shuttered. Or perhaps turned into vocational colleges if there are still some jobs humans need to be trained for.

But even if you accept top-tier unis can continue to operate as finishing schools for the most elite and/or ambitious members of the professional-managerial class, you have to wonder why anybody will spend a big chunk of their precious youth learning to be, say, a lawyer or accountant or engineer when AI is eating all the lawyer, accountant and engineer jobs.

The collapse of society
There’s now a bipartisan consensus that the Anglosphere nations handled the deindustrialisation poorly. Millions of blue-collar workers were thrown on the scrap heap, with many of them proceeding to commit slow-motion suicide or the more abrupt variety.

Most former miners and manufacturing workers tended to turn their anger and despair inward. I don’t imagine their professional-managerial class counterparts will be so inclined to self-blame. As my fellow Substacker Peter Turchin frequently observes, if a society has more elite aspirants than elite positions, it’s headed for trouble.

Late Rome had an oversupply of nobles. Pre-1789 France had a surplus of lawyers and intellectuals. Pre-1917 Russia had more university-educated individuals than the monarchy or church could find jobs for. If there is a white-collar bloodbath, first-world societies will soon have a surfeit of pissed-off professionals with plenty of time on their hands.

Let’s hope for the best but start planning for the worst
It’s worth noting that the benefits of the Industrial Revolution weren’t redistributed without a hell of a fight. From the moment the gentlest of labour laws were introduced, mill owners reacted furiously, warning that such initiatives would collapse productivity and profits, fatally undermine property rights and the free-enterprise system, and ineluctably lead to socialism. The children of those mill owners fought the nascent trade union movement, just as their grandchildren combated moves to extend the franchise to the working class. Their great-grandchildren also had plenty to say about post-war Keynesianism and the welfare state. Plus ça change…

Maybe my years in journalism have made me cynical. But I don’t expect the modern-day mill owners – the Altmans, Andresseens and Jassys of the world – to be any more inclined to share the (AI-generated) wealth around than their Industrial Revolution predecessors were.

If you don’t want to be reduced to neo-Dickensian poverty levels between now and 2030, dear reader, you might want to consider politically mobilising while there’s still time to shape events.