In 1930, John Maynard Keynes predicted that within a century, technological progress would reduce the working week to fifteen hours. He was right about the productivity gains. He was wrong about the hours. The surplus went somewhere — just not to workers.
When washing machines and dishwashers arrived in American homes, the same pattern played out at the household scale. Historian Ruth Schwartz Cowan documented what actually happened: the hours didn’t shrink — they redistributed. Clothes washed weekly got washed daily. Standards rose to fill whatever space the technology had cleared.
The system absorbs the surplus.
LLMs follow the same logic. The open-weight models — the ones you can run locally, for free, right now — are already capable enough to fundamentally alter the experience of work. Development could stop tomorrow and the transformation would continue anyway. The box is open.
The Wall Street framing — fewer humans, same output — is too simple. Rising expectations will pull LLM-equipped workers into new gaps, change the shape of the work rather than eliminate it entirely. This is Neutral or positive news for the labour market (depending on the shape of those gaps).
Which makes this moment the good part. The brief moment where the productivity gains are real and the expectations haven’t caught up. The question is where the surplus goes. The system will absorb it.
The people who are refusing that are making bets: on using the tools to multiply what’s uniquely theirs rather than replace it, on relationships and the communication skills that make them durable, on a particular kind of attention — the ability to see the whole picture while holding the details in view. None of these are easy bets. But this is the window in which they’re cheapest to make.