OpenAI published a policy paper this week proposing robot taxes, a public wealth fund, and a four-day work week as responses to AI-driven displacement. Before criticising it, I want to acknowledge what it is doing: most of the AI industry is still insisting that displaced workers will retrain, that new roles will emerge, that the historical pattern will hold. OpenAI is saying something more honest. Something structural is happening, it will be large, and it requires a policy response. That is worth crediting from a company with every commercial incentive to say otherwise.
The problem is that the serious thinkers who have modelled this trajectory arrive at numbers that make OpenAI's proposals look like gestures. When you work through what happens to employment when machines become better, faster, cheaper, and safer than humans at all tasks, the central estimate is not a marginal contraction. Labour force participation collapsing from 62% today to somewhere between 10 and 20% is where the modelling consistently lands. Not as a worst case. As the most likely outcome. A 75% reduction in employment, concentrated over a generation or less.
Both OpenAI and the post-labour economists who have done this work share a common conclusion: in a world where wage labour is no longer viable for most people, the answer is capital ownership. People need to own a stake in the machines. Receive dividends from their output. Be connected to the productive economy through ownership rather than employment. As a description of a functioning end state, this is coherent. As a path from here to there, it is close to fantasy for the majority of people currently living paycheck to paycheck.
Capital does not redistribute itself. The people most exposed to displacement, warehouse workers, call centre staff, junior knowledge workers, have no savings, no investments, and no political leverage to extract ownership stakes from the companies replacing them. A sovereign wealth fund or a public dividend requires the state to acquire those stakes before the value is fully captured by the people who already own things. That is a large, contested, politically difficult project that neither OpenAI's paper nor the post-labour framework spends much time on. Naming the destination is not the same as describing the road.
But there is a deeper failure in both frameworks, and it is not about economics at all. Neither seriously engages with what mass unemployment does to people beyond the loss of income. Work is not just a pay cheque. For most people it is identity, structure, daily purpose, and the primary way they locate themselves in relation to other people. The meaning crisis that researchers have been tracking for years, rising rates of depression, social isolation, deaths of despair, is already partly driven by the erosion of stable employment in communities where industries have collapsed. We have not seen anything yet. A 75% reduction in labour force participation is not a policy problem with a redistribution solution. It is a civilisational rupture, and the psychological and social consequences of that rupture are not addressed by a wealth fund, however well designed.
The honest thing to say is that we do not yet have good answers to what a society looks like when most people are economically optional. The post-labour economists have modelled the labour market. They have not modelled what replaces the organising function that work has performed in human life for all of recorded history. OpenAI's paper does not get close to that question. Neither does most of the serious thinking in this space. The problem deserves better, but to get better answers we first have to be honest about how inadequate the current ones are, including the ones being offered by the people who correctly understand the scale of what is coming.