The end of labour

We are living through the final dawn of human labour. When we achieve AGI capital will be the only means to accumulate more capital. Machines will be so much more effective in all walks of life that the value of human labour will be trivial. We are living in a golden era where AI can leverage labour to disrupt entrenched capital in a way never seen before, but this era will end and when it does society will become stuck, with no possibility of individuals improving their (relative) material situation.

Before, a few things that won’t be covered.

1. If/when AGI is going to arrive. This (dilettante) author believes we are no closer (technically) to AGI than we were 5 years ago. An artificial general intelligence would, due to the scaling of bits over atoms, only need to be as smart as the average human to trigger ‘the singularity’ (the moment AI is incomparably more intelligent than humans). Given the difference in processing between a human brain and a monkey is only a few petaflops (far smaller than the current largest AI models), it would stand to reason that our failure to create an AGI is not due to lack of computation but a fundamental misunderstanding of how human intelligence works. This does not however mean we are no closer to AGI (if AGI will be created then of course, we can’t not be closer than we were yesterday). Trillions of dollars are flowing into AI investment, GPT, while perhaps not smart, has given people a glimpse of of what is possible. We’ve stepped through the looking glass and the road to AGI certainly feels inevitable.

   
   

2. The UBI question. Some particularly optimistic proponents of AGI will argue that capital will be available in such vast reserves that, like a Norway on biblical levels of exogenous hormones, no one will want for anything materially. To this I answer, human nature knows no limit, our fallible brains have evolved over millennia to see success only in relative terms, there will always be a need to have more than Sally next door and markets will always find a way to provide this need. It’s futile to deny our conditioning.

Labour vs Capital

In our capitalist world, there are two ways to produce capital. Via labour, or through the breeding of capital itself.

Before we dive deeper, we should quickly define what is meant by labour and capital. I’m going to define labour as any type of useful human work. This could be as tangible as the Andaluza jornalera plucking a sun-blemished tomato, or as abstract as the anodyne city intern, perspiring over the row widths of her latest pivot table.

Capital is anything non-human that can be exploited to produce more capital. I tend to imagine capital as a self-mutilating hydra, relieving itself of one head in order to sprout two more. For readers with less flowery inclinations, picture a tractor, a computer, a data centre or money. I should make clear here that capital is not just money, indeed money is only capital in the sense that it can be liquidated to acquire ‘real’ capital (that can produce more capital). Money not exchanged for useful capital will, over enough time, be eroded to nothing.

Over time the ability of labour as a mechanism to produce capital has (barring a few exceptions we’ll cover soon) fallen fairly dramatically vs the ability of capital itself. The easiest way to visualise this is through comparing productivity to wages. Productivity is the output (as capital) per input block of human work, wages are what this worker gets paid. In a world where labour and capital were equally good options to generate capital, you’d expect to see wage growth increase in proportion to productivity – If a Northampton cordwainer increased his output of loafers, derbies and brogues using the latest CNC machining technologies by 50%, his wages should rise 50% also.

Of course, this is not what has happened. Wage growth has remained flat over the last 40 years, while productivity has increased 3 fold. The reason is simple, the capital has flowed to those that own the capital of machines and the distribution network (shareholders), not to the workers trading their labour for capital.

Do the rich get richer?

Anna is a wagey with a protestant work ethic, and her friend, Alex, a part-time designer (with a more catholic work ethic) but with a distant, wealthy uncle. Anna has spent years building up ‘market value’ in a high-demand niche, say, analysing liquidity risks for commodities hedge funds, she now sells her labour for £150k/year. Alex works a couple of days a week and takes home £30k/year.

After tax, Anna takes £80k, Alex £25k. Anna has (miraculously) avoided the hedonic treadmill and manages to save £55k/year, Alex just manages to break even on his lower salary, but doesn’t save a penny.

Then – a black swan. On a shoot at his Perthshire estate, Alex’s uncle catches a stray from his cousin's Remington 870 pump-action and (tragically) passes away, leaving Alex the large, but not exorbitant sum of £1m. Alex invests this and receives a standard market return of 10% annually, his capital grows to £2m in 8 years. With Anna, even increasing her savings by 5% each year and also getting a return of 10%, it takes double this time. This is the modern-day power of capital.

This example may paint a fairly bleak picture. The rich get richer and the poor poorer, but it doesn’t quite line up with the facts. Only 1/3rd of the billionaires in America inherited their wealth, the vast majority of the new ultra-wealthy then are indeed creating capital from labour. How is this possible when we’ve so clearly shown that the return of productivity goes to the one who holds capital?

The Exception

Entrepreneurship is one of the few, if not the only, paths where labour can create capital more effectively than capital can. Entrepreneurship has shown time and time again that when it comes to disrupting capital, holding capital does not count for much.

This year Tesla sold 1.8m electric vehicles. VW sold just 800k. In 2003 Tesla raised $6.5m in it’s first proper funding round, at the same time VW had cash equivalents of $12bn (1800x more), they were actively working on electric cars and had 20 years prior, in 1981, released the Golf I CitySTROMer, their first fully-electric prototype.

From a capital point of view, this doesn’t make much sense. VW were ahead of the EV race in every conceivable measure, the difference is we (for now) live in a world where human ingenuity counts for something. Combining the right people with unwavering focus and a mindset to question absolutely everything about the status quo of the industry, and of course, an amount of luck, can mutate a tiny amount of capital to create outlier outcomes.

Entrepreneurship then, is the solvent that cuts through the resin of entrenched capital, buoying new winners while keeping the existing honest.

The End of Entrepreneurship

Generally then, capital beats labour, but not always. An entrepreneur operating with a high degree of focus, ingenuity and in a system with a lot of randomness, can achieve outlier success. So what happens when entrepreneurship is made redundant? The first obvious thing to me is that if this labour class is made redundant so will all other classes (apart from perhaps one or two we’ll get to later). If Marc Andreessen can bypass investing in founders by instead investing in AI agents, then you can bet that the VP of your local bank is going to bypass paying barely-pubescent McKinsey consultants $5000/day.

Some may think it's sensational that you’d ever have an economy where capital only flows to AI agents, but I don’t think so at all, it’s not just plausible, it’s inevitable. AI agents will communicate quicker, deploy code more efficiently, leverage and coordinate humanoids more efficiently. They’ll know us more deeply than we know ourselves so will be able to craft narratives and manipulate us to buy their product in ways even the best marketers today cannot. As soon as a promising avenue is identified it will be able to be scaled up in hours, ventures that don’t show promise will be dispassionately discarded and replaced with a new one as quickly as you can fold and be dealt a new hand. Co-founder conflict is currently the number three cause of startups failing (after no market and running out of money), this will just be a non-issue for AI agents.

Perhaps it’s not to say that entrepreneurship will end, it might be more accurate to say we will see an entrepreneurial renaissance, companies we can’t even dream of will be created in weeks. It’s just that these companies will operate completely devoid of any human labour. The only thing that will matter is capital, it will make the chart we show on wage vs productivity growth earlier look comparatively egalitarian. Capital will birth capital and labour will birth nothing, social mobility will pretty much cease to exist, society will play out like your favourite Shakespearian comedy. Enjoyable, but with no doubt in your mind about how it will end.

We’re only human

Post-AGI there will likely be one class of work that facilitate some kind of social mobility. This is labour where it’s value is to deeply intertwined with the labourer themselves being human, it will, by definition, be impossible for AI to replace. Just as the advent of specific AI’s has done nothing to quell the thirst for competitive chess, there will be other last bastions for human ingenuity to seek refuge.

Luxury goods should remain an industry safe from AI. After the Singularity an AI agent could spin up a luxury Swiss Horology startup, it could craft the most elegant timepieces imaginable, a perpetual calendar in a case 1mm thick, a split second chronograph, accurate to 1/1000th of a second, bezels and dials crafted from the most innovative materials. There will, likely be demand for such a product, but aficionados will still desire timepieces (painstakingly crafted by the trained hands of a horologist). It’s also difficult to imagine the status derived from the brands most steeped in history will disappear in an AGI world. The same will be true of art. AGI will (and indeed AI already does) create new markets for machine-created artwork, but enjoying art is not simply a passive act of admiring aesthetics, it’s an active experience of a shared human experience, between you, and the artist. Proust put it best ‘Through art alone are we able to emerge from ourselves, to know what another person sees of a universe which is not the same as our own and whose landscapes would remain as unknown as those of any planet but for the art which reveals them to us.’ As long as machines are othered, artisans and artists will have a place.

The final labour type that it’s difficult to see being dislodged is what David Graber would call Social Labour. Nurses, doctors, coaches, waiters, barbers. This is labour where, on first inspection, it might appear you are paying for expertise, but digging deeper you realise a large part of their value comes from the social connection you receive. Having a chat with your barber you’ve been coming to for years, eating at your local Sicilian restaurant and having the owner's daughter talk you through the family recipes, these are all luxurious experiences that will still be highly desirable.

The last dance of labour

We’re closing in on the endgame of AI, and in many ways, we’ll never have it better than we have it today. Right now, AI is a current amplifying our labour. It can help us start businesses, spread ideas, clarify thoughts and organise people in a way we’ve never been able to before, it’s never been easier to do more with less. But this will end. It’s like watching an avalanche that is most turbulent before it settles. As soon as the endgame finishes, it’s hard to see how, apart from some of the niche roles we’ve described, labour will ever have a fighting chance against capital. Now is the time to build.