In another age, when Rishi Sunak was still the British prime minister, and Elon Musk had not completed his journey to become a full blown Magalomaniac, Musk suggested to Sunak that AI would make all “work” unnecessary. Sunak wanted to argue that work was important for our sense of purpose and self-esteem. Musk agreed, in effect, that we needed to make decisions about what we wanted.

Any discussion about AI quickly turns to the question of whether it can replace human capability. The sceptics doubt that. The evangelists (including, most recently, Bill Gates) claim it will surpass human capability for “most tasks” in the not-too-distant future. I suspect the evangelists are not sufficiently critical about what they mean by “surpass”, but don’t doubt that the coming wave of automation will go wider and deeper than anything we have seen before.

Truly disturbing here is the absence of any serious public discussion about the effects of this on our social structures, our politics and economics, and in the most profound way in what we value in human endeavour and our lives. Without that discussion, that articulation of what’s at stake, we are hardly going to be placed to make the decisions Musk (with uncharacteristic apparent modesty) rightly observed we need to make.

If the tech evangelists are right, the opportunity for most people to work could largely disappear (or at least be radically diminished). The underlying assumption is that automated processes will be more economically productive than human processes, and so there will be no reduction in the amount of value generated in the economy. Indeed that value should increase.

This productivity is important in western societies, where ageing populations have resulted in a top-heavy stratum of pensioners, who might feel with justice that they have paid their dues through their working lives, but where the economic sustainability of their pension funds depends on the productivity of the current working population. The relative size of the working population has shrunk and will continue to shrink given the current trend of falling birth rates: the working population needs to generate more economic value from less work.

If that value is being generated by non-human agents, there is an immediate and revolutionary question about its distribution. If the value moves simply to the “owners” of the productive assets (the AIs), who are already wealthy enough to hold many non-productive assets, then we really are looking at a new feudalism, with the difference that the feudal overlords will have no specific need of their serfs.

But then economic collapse looms, because in reality the overlords would depend, not on their serfs’ labour, but on their spending. If the serfs have no money, then AI productivity becomes pointless, because too few people can buy its products. Economies are fundamentally circular, dependent not just on overall productivity, but on how the money moves through society.

The academic and tech investor Scott Galloway has argued that AI tech will not exactly function this way, quoting the FT columnist Robert Armstrong.

“There is a new vision where [AI] is much more competitive, and profits are shared, and much of the value may be captured by consumers.”

Galloway and Armstrong were responding to the market impact of the Chinese upstart DeepSeek. DeepSeek may not become the default AI tool, but it has already changed wider perceptions of what’s going on, and raised the likelihood that AI technology will be commoditised, becoming a general feature of more efficient business and industry rather than channelling available profit to the AI overlords.

This still leaves the question of the forms in which AI might deliver value, because we’d still be looking at a world where profit was generated in diverse industries and sectors without much human input. In the normal way of things this profit would mostly flow to the “owners” of those business assets, leaving the necessary question about what will happen to all the displaced workers.

The scale of this displacement means that new roles look unlikely to fill the gaps. Silicon Valley types are given to claim that their technology will free people to concentrate on more interesting or higher value tasks, but you have to wonder what those tasks might be if these evangelists are also correct in their claim that AI capabilities will surpass human ones for “most tasks”.

In theory a major benefit of AI is the distribution of greater leisure to everyone, but if it’s going to happen we have to answer how enough of the generated economic value will be diverted to the general population.

Perhaps it could start in some form of a Universal Basic Income (UBI), but the viability of a UBI system really requires nothing less than a fiscal revolution.

It would also need a cultural one, in which it was more than acceptable for most people to be “living off benefits”, at least for their basic needs, and probably more than the basics.

In this extreme version of the future, most people will not need to work (as Musk claimed). But even if this proved possible, for many, if not most, work means more than financial subsistence. As Rishi Sunak argued, it becomes a ready source of purpose, if not of meaning. If that’s no longer available, what are people going to do with themselves?

Culturally we see work as a moral quantity. For much of the classical world work was something which only inferior people had to do, and consequently was looked down on, a view that persisted into the modern world particularly among the aristocracy. With the rise of the middle class (and perhaps too the influence of the US myths of self-sufficiency) things changed. We’ve come to equate work with merit. Generally we say people who work hard deserve to be rewarded. Those who live off benefits, for whatever reason, are routinely seen as scroungers, as the undeserving poor. Conversely the rich are usually seen as somehow having earned their wealth, though much of the time this is barely true.

Right wing politicians and their supporters are fixed in this attitude, and forever see themselves as the paying victims of people less deserving than themselves. But the left is not immune to this thinking either. Though they have a better understanding of the point of safety nets, and how they can reasonably function, they will still talk of a right to work, and frame their economic arguments with an eye on its desirability. They will also target the concept of “unearned income” as a starting point for redistributive efforts. Britain’s Labour government will routinely speak of “hard working people” as its core constituency, leaning on deep rooted notions of merit and fairness. In its current right-shifting form, that government shows no sign of even considering that there might be a problem here.

We need to be more thoughtful about these assumptions, even if in all likelihood things will be less extreme. AI will certainly displace jobs, but in an optimistic reading will do so in a way that allows us to recalibrate how work functions in our societies, most obviously through a marked reduction in the standard working week.

There will be plenty of managerial resistance along the way: the pressure post-Covid for employees to return to the office demonstrates how entrenched conventional ideas of employment remain. The few brave businesses who have already moved to a four day week have consistently reported good outcomes and yet they remain in a tiny minority, as if there was something naturally ordained about the current order of things. You don’t need to know much history to understand that the 40 hour working week spread over five days is relatively recent, and achieved through worker pressure on reluctant managers. It changed before and it can change again. There’s certainly nothing intrinsically necessary or even efficient about it.

If working people have an extra day or more’s leisure this will in itself have far reaching consequences. The redirection of wealth from the majority to the already rich has entrenched the current expectation that people need to work long hours, perhaps even in multiple jobs, simply to make ends meet. A redistribution of the gains from AI could and should reverse this trend. It should also end the stupid correlation of long hours with good or important work.

It should allow us as a society to think more intelligently about the distribution of parenting responsibilities within a couple, and the effects of properly shared parenting on career prospects.

If people have more leisure it’s likely that they will start to want new and different ways of enjoying it, which could in itself become a source for new employment, facilitating those leisure activities, so you’ll see some job creation as a direct result.

It could improve social mobility by helping people develop new skills and capabilities in their time away from their primary work, potentially enriching other lives as well as their own.

All of these things could be positive and useful, but nothing will happen without radical thought about how the necessary redistribution of national wealth could be managed. Traditionally that would be done through some form of taxation, but taxation has proved a problematic instrument. Most people resent it, and the wealthy have become adept at avoiding a proportionate payment of their income (apologists for the status quo routinely claim that the wealthy already pay the larger part of the tax take, which may be true but conveniently ignores the force of “proportionate”).

At the very least we need to reframe fiscal systems, so people understand that they do not represent the government taking an individual’s “hard earned money”, but are a necessary intervention in the circulation of national wealth.

This idea, cutting against centuries of resentment, is hardly going to be an easy sell even if it’s true, but then we’re moving towards a world where we become clearer eyed about the non-relationship between how we spend our days and our disposable income, for everyone rather than the wealthiest and the poorest.

We need a radically new fiscal system fit for this world, built around the concept of fairness, because a rigorous idea of fairness could be the only starting point for such a system. This might seem Utopian, but it proceeds from the recognition of a reality that will have to be dealt with.

Not least we will need to rethink education, partly so people are better equipped to cope with more leisure (and to find their own purpose in it), partly to do the groundwork for the new skills people will need in this brave new world.

Coping better with leisure might sound like a nice problem, but it is a real issue. In my first professional job I produced a staff magazine for a telecoms manufacturing company. Part of the job entailed interviewing people as they were about to retire, and I’d routinely ask what they were planning to do with their extra time. Almost invariably, at whatever level they’d worked in the business, the answer would be “a bit more gardening”.

While there’s nothing wrong with gardening, if you like that kind of thing, it’s not likely to be enough to fill the chasm created by the loss of full time purposeful work. We need to help people deal with a shift from finding meaning, or at least purpose, in their work, to finding meaning more broadly in what they do.

People will need new skills, and the core flexibility to adapt to whatever new types of work might be created. With political institutions struggling in the face of facile populism, we also need a clearer focus on the critical thinking skills required to filter the garbage of half truths and lies that fill the internet, garbage likely to be generated in ever greater quantities by an unthinking use of AI.

If mishandled AI itself could contribute to the loss of those critical thinking skills, by encouraging us to outsource the valuable hard work of research and conceptual development. There’s a lazy misconception that by letting AI do the spadework we can concentrate on some higher layer of creative thinking and insight, but that really is like asking someone to advance the intonation of their instrumental performance without first learning to play the instrument. The skills required by any further layer of insight are developed by the spadework.

None of this is going to be easy, but if AI develops as its evangelists are suggesting, nor is it optional. It’s a challenge that has to be addressed first at a political level, through a clear and coherent political vision, and right now, with a rogue version of the United States demolishing the post war international settlement of the 20th century, we could fairly say that the world’s politicians already have their hands full.

But these questions about the future of work and its relation to societal well-being will be critical in shaping what comes next, and however difficult it might seem, we have the power to make the right choices. We need to exercise that power, starting with a proper recognition of what we’re up against. Through that recognition we might even develop a political vision to counter the idiocy of populism.

AI will change everything, but left to itself will not save us. That’s the reality Keir Starmer and other politicians desperately need to address. For the moment, for the most part it seems to be way beyond their managerial imaginations. We have to demand something better.