A prepper’s guide to the looming AI tsunami

Some thoughts on how the next few years are likely to play out

Photo by Gerard Siderius on Unsplash

Video summary

Soon, new intelligence will be able to do anything humans currently do on computers — including every data processing or “knowledge” job, managing employees who do not work exclusively with screens and keyboards, and even running entire companies. We need to start preparing for that moment… even without any fundamental technological breakthroughs, if companies just keep training LLMs on more domains, enlarging the context windows and making more sophisticated agents, then new intelligence will soon be capable of replacing up to 40% of workers in rich countries.
Zack Exley, New Consensus, 4/2/25

By the end of 2025, there is no technical bottleneck to… most legal work being done entirely by AI. However, diffusion takes time. Also, in many countries lawyers barricade themselves behind a cluster of laws that forbid lawyer-automating AI. Getting hired as a new lawyer, or any kind of white-collar analyst, is getting harder though, as decision makers expect AI to reduce their need for entry-level white-collar workers of every kind, and firing people is much harder than not hiring them in the first place… AI doctor appointments are only trialled at any significant scale in 2026, by Singapore and Estonia. Significant integration of AI in the non-patient-facing parts of the healthcare system is underway in the UK, many EU countries, South Korea, and China by 2026, but again diffusion is slowed by the speed of human bureaucracy.
L Rudolf L, No Set Gauge, 17/2/25

But CEOs are looking at what Musk did to Twitter, and what he’s doing to the government, and thinking, “why not me?” A host of corporate leaders are now trying to cut their workforce and replace it with AI. In technology, Microsoft is firing people, Meta is getting rid of 5% of its workforce, and HP and Salesforce are following. Layoffs are happening at Google, Autodesk, Blue Origin, Sonos, eBay, Sophos and Intel. In the non-tech economy, UnitedHealth Group’s Optum, Starbucks, and Walgreens, among others, are engaged in mass layoffs. Stanford and Harvard are freezing hiring, as are higher educational institutions across the country.
Matt Stoller, BIG, 3/3/25

Inconveniently, Lenin never actually said, “There are decades where nothing happens, and there are weeks where decades happen.” But if he had said it, it would be an apt description of where we find ourselves at the quarter-century mark of the new century.

What’s left of the fiscally conservative, socially liberal consensus of the last 45 years is being obliterated by Trump 2.0. The post-WWII rules-based liberal international order is being, well, re-ordered. Not unrelated to those developments, the once all-conquering Professional-Managerial Class (PMC) is facing a vociferous backlash after spending at least a decade massively overreaching.

Given the unfolding political realignment, aka ‘vibe shift’, forecasting even the near future is complicated enough. Complicating forecasting further is the much-heralded but little-understood arrival of the ultimate game-changing technology.

2025 is when the AI rubber starts hitting the road
At this point, I’m almost contractually obliged to bust out the relevant I.J. Good quote, so here goes:

Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an 'intelligence explosion,' and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make.

To dumb Good’s point right down: once humans invent a technology smarter than humans, there’s little need for humans.

I believe we are careening towards that intelligence explosion tipping point and that the AI job apocalypse is nigh. I’m hardly a lone voice in the wilderness, but my sense is that most people are still oblivious about how hard and fast the AI-driven wave is about to hit.

As someone who lived through the collapse of the print media and who has seen Gen AI hoover up a lot of content-creation work once done by humans, I may be overgeneralising from the specific examples I’m most familiar with.

But more and more experts, disinterested and otherwise, now appear to be reaching the same conclusion as me – the logic of capitalism, especially the devil-take-the-hindmost iteration of it that’s been hegemonic since circa 1980, means businesses will use AI to automate everything they can and start jettisoning their human workers as soon as it is feasible for them to do so.

I imagine we’ll soon have hockey stick unemployment graphs in first-world nations. A handful of early adopting companies will automate away their human workforce, and competitors will need to follow suit if they are to remain competitive. Alternatively, AI-leveraging start-ups will drive old-school, human-employing companies out of the market.

With more or less delicacy, the likes of Goldman Sachs, the IMF and Brookings have all rung the warning bell about profound labour market disruption occurring before 2030. Techno-optimist Substacker James Pethokouis recently referenced a JP Morgan report that predicts 2025 will be the “Year of Agentic AI”, 2026 will be the “Year of Physical AI” (i.e. robots), and 2027 will be the “Year of Artificial Super Intelligence”.

Let me put that in terms even the technophobic can understand:

1) If you’re a ‘brain’ worker, you can expect your skill set to become obsolescent this year

2) If you’re a ‘brawn’ or ‘brawn and brain’ worker, that process will kick off in earnest next year

3) By 2027, it won’t matter much what you once did for a crust because Artificial General Intelligence (AGI), or possibly even its more accomplished sibling Artificial Super Intelligence (ASI), will be able to do it vastly better for vastly less than your employer or clients currently pay you

Mere AGI just replicates human intelligence, so it will only lead to enormous economic and social changes. ASI far exceeds even the most brilliant human minds. It will lead to presently unimaginable cultural, economic and societal transformations.

The JP Morgan analysts could be off with the timing, as could all the other venerable institutions and accomplished individuals suggesting that AGI, and maybe even ASI, is only a few years away. But they are almost certainly in the right ballpark. If you’re not doing a different ‘job’ and living in a much-transformed society in 2027, you almost certainly will be by 2030.

OK then, 2035 at the absolute latest.

Look at it this way – preparing for the AI apocalypse a little early has few costs and many potential benefits. But leave things too late, and your options will be considerably narrowed.

What is to be done?
Lenin did write a famous political pamphlet entitled, ‘What Is To Be Done?’ and that is, as always, the question.

If you believe that AI is being hugely overhyped, mainly by those who stand to benefit from its rapid adoption, the good news is that you don’t have to do anything. After all, if the world isn’t about to change profoundly, you don’t need to course correct.

But what if you’re not so sanguine? What if you suspect we’re about to experience something similar to the take-up of the internet over the last three decades, telescoped into about three years?

In this week’s post, I’ll attempt to briefly sketch out the labour market changes and political responses that will occur once employers get in touch with their Inner Musk. Next week, I’ll take more of a ‘micro’ perspective and provide some advice about remaining economically viable for as long as is feasible in an economy where machines almost always outcompete humans.

First they came for the Professional-Managerial Class
As riven with class conflict as human societies have been for the last 6000 years, the classes have hitherto needed each other. To put it bluntly, the elite class has provided the brains while the masses have provided the brawn, serving as farm/factory/cannon/office cubicle fodder.

AI means the masses are no longer necessary.

AI ultimately means the elites aren’t necessary either. But that’s a whole other Musing, so let’s focus on what’s about to happen to the workers – including late capitalism’s PMC labour aristocracy – academics, accountants, engineers, lawyers, management consultants, marketers, etc.

Get ready for the anti-automation laws
Frankly, I’m surprised there hasn’t been more public discussion about introducing such laws already.

L Rudolf L, of No Set Gauge, predicts some nations will simply declare themselves ‘human only’ and outlaw the use of AI in the coming years. Rather more common, I suspect, will be make-work schemes that require a human to sign off on what an AI has done. All parties involved – the worker, the worker’s social circle, the worker’s employer, the AI – will understand this is all human-ego-salving theatre, and that no oversight is required. But pointing this out will be considered extremely rude. (Much in the same manner it’s been verboten to notice any of the downsides of vibrant diversity in recent times.)

Those professionals with strong unions and/or potent lobbying capabilities – think doctors, pharmacists and lawyers – will likely have many early political victories in sheltering themselves from the AI automation threat.

What democratically elected politician isn’t going to wilt in the face of an emotive ad campaign featuring, I’d guess, a sinister-looking robot flatly informing a tearful mother in a hospital bed, “Your medical tests are back, human. You have approximately 2.5 weeks to live. Please settle your account prior to this date and have a pleasant day.”

So, expect those with clout to at least slow, if not outright stop, AI automation for a time. Those without political muscle – data entry clerks, for example – will be granted no such stay of execution.

But any reprieves the well-connected win for themselves will likely be temporary. The average voter may be warmly inclined towards their avuncular family doctor and friendly neighbourhood pharmacist. But they will inevitably come to wonder why there is a two-tier approach to industry disruption. One where a handful of favoured industries are sheltered from the AI automation wave while many others must reckon with Schumpeterian creative destruction.

As an aside, while physicians and pharmacists might get some support from the general public, other members of the professional-managerial class would be optimistic to expect any. After spending the past four decades throwing the common folk under the bus economically while sneering at them culturally, I fear the Anglosphere’s academics, artists, media workers, public intellectuals, senior public servants, scientists and software engineers are about to learn all about the downsides of being a bad ally.

The ’protected species’ issue will burn all the hotter when national governments less inclined to indulge special interests – L Rudolf L is probably on the money in predicting this will be a Singaporean or Estonian government – start allowing AIs to replacing human doctors, pharmacists and lawyers.

Given ageing populations and soaring medical costs, it’s unlikely any Anglosphere nation will long delay the automation of tasks now performed by (well-remunerated) human doctors and pharmacists. Especially once the example of Singapore/Estonia shows AI doctors and pharmacists are at least as competent and vastly cheaper than their human equivalents.

Lawyers are massively overrepresented among the political class but not widely beloved by voters. So, I don’t imagine our learned friends will fare any better than their healthcare-professional mates in maintaining restrictive work practices.

Much of the work currently performed by humans will be automated away over the next five years, and pretty much all will be automated by 2035. But as in every other historical period, there will be winners and losers.

Next week, I’ll aim to provide some tips about how people can keep ‘winning’, at least in the sense of being able to generate an income, over the next couple of years.