A real-world trial for work in the age of AI

Previous technological disruptions have destroyed “jobs”. But they have not destroyed “work”, because they have created at least as many employment opportunities as redundancies — in many cases, far more. Perhaps artificial intelligence will end up the same way. Or maybe not.

The remarkable leaps it is making may mean that even jobs we don’t think of as at risk from AI may in fact also be in danger. If AI can accelerate the rate of advance in robotic research, occupations that we currently think of as “AI proof” may also soon be taken by machines.

One response — articulated last week by Bill Gates — is that we should have jobs that are “human reserved” and cannot be filled by robots. Another is that we should all transition to a life of leisure, with our incomes provided through cash transfers and a universal basic income, or UBI.

People often talk as if UBI is largely untested, or confined to a handful of flawed trials. And it’s true that most “universal basic income” trials are a lot like the Holy Roman Empire, which was neither holy, Roman nor an empire — they are neither universal, basic, nor, in some cases, an income.

There is, however, a real-world trial that has been running in countries across the developed world and tells us something about what a world with UBI would look like. It is called the state pension: an essentially unconditional cash transfer to retirement-age adults.

Let’s set aside for a moment the fact that there are pensioners who, because they do not have enough to get by or do not have security of housing tenure, are forced to work. Such cases don’t tell us very much. What’s immediately striking when you look at state pension schemes is that the image often conjured up when we talk about UBI — of people sitting around enjoying the government’s largesse — is mostly wrong. Take the UK, for instance, where many choose to work well beyond pension age. A minority continue in paid occupations that they enjoy, and others might take up paid work as a way to meet people.

But when we expand our definition of “work” to include voluntary labour of the kind that we would otherwise have to pay someone to do — childcare or providing “social infrastructure” by running an apartment block’s homeowners’ association, serving as a magistrate, volunteering as a tour guide at a museum or a gallery — things look rather different.

Of those who don’t do this, of course, some have delightful lives filled solely with leisure: they might go on long holidays, spend a lot of time at the cinema or concerts, read or play board games. But then there are those non-working pensioners who spend their time going down online rabbit holes becoming supporters of extreme ideologies of one kind or another.

What the state pension experience also shows, then, is that when voluntary work dries up, that takes a psychological toll just as the loss of paid work does.

This tells us something important: a more generous welfare system might become an important part of ensuring that most of us have a decent standard of living in a world of AI and advanced robotics. Paid employment might dwindle significantly, but for most people some kind of “work” is going to be an important part of their sense of wellbeing. If my job is replaced by a machine, I will need to replace the lost income. But I will also need some work to do, whether it is paid or not.

It all comes with greater risks for the young: although those of us who are middle-aged or older are not immune to extreme ideologies, the young are more likely to turn to violent means in pursuit of them. That some pensioners turn to extreme politics in retirement is alarming enough — but as a gauge of what people at the start of their lives might do without work, it is actively worrying.

This is why Gates is right to advocate for some “human reserved” roles. Most of us who support minimum wages and some restrictions on the right of employers to fire at will know that these policies come with an economic cost. But we judge that the economic cost is worth the social benefit of a wage floor and some workplace restrictions.

Judging the right balance is the stuff that most politics is about. And so in the age of AI, much of politics will be about figuring out the balance between the economic benefit of letting a machine do the work and the social benefit that most of us gain from working.

[email protected]