Sunday 28 April 2019

What Kind of Jobs Will the Robots Leave Us?

Coming for your job!
Machines powered by self-learning algorithms and internet connections are displacing humans from all kinds of jobs, from driving to legal discovery to acting in movies. Will there be any work left for us to do? Economics says yes. Will it be awful or will it be nice? That is up to us.

I. The Paradox of Automation

We have two hundred years of historical experience with technological revolutions that rendered vast numbers of jobs unnecessary. In every case, we find a paradox. On the one hand, human labour not only survives such revolutions but even becomes better paid. On the other hand, the economic value of the work we do in our new jobs falls. We get paid more to do things that don’t matter as much.

For example, in 1800 the overwhelming majority of people still worked on the land, even in Adam Smith’s Britain. They had to. Producing food required enormous amounts of human labour. Most people spent most of their income on bread, which was also the main product of the economy and the obsession of classical economists.

Then came the agricultural revolution. New technologies and managerial methods spread and productivity per agricultural worker soared. Millions of agricultural workers were no longer required - yet they did not starve. They were available to do other kinds of work, making things in the new factories and living in the new towns that sprang up around them. They went from growing food (essential to human survival) to making things like cotton shirts that were merely nice to have (helpful to human living).

Then the same thing happened again in the factories. They became more efficient at making things. Output rose. Millions of jobs were made obsolete. Millions of people found new work performing helpful services for each other. In economies like the USA, furthest along this trajectory, farming now makes up less than 1% of employment and economic activity; while services make up 80%. (By 2000, more people in the UK were employed making and selling sandwiches than in the entire agriculture sector.) Obviously governments can mismanage the transition – millions of young people across Africa, the Middle-East and India can’t find jobs. But not because of some law of economics.

The answer to the paradox is this. Technologies like automation increase productivity, the amount of economic value a society can produce with the same inputs of labour and raw materials. That means we can have all the things we used to have, plus we now also have some spare labour left over. We can use that spare labour to produce things lower down on our collective list of priorities, such as mass higher education and healthcare and interior decorating and pizza delivery. The reason we didn’t make those things already (or not much of them) is that they were not valuable enough to give up anything higher on the list.

The same economic logic applies to the current algorithmic revolution in productivity. Robots can only take our jobs by making our society richer. Many hundreds of millions of jobs will be eliminated, but that does not mean that paid work itself will disappear. There will always be more things that people want doing for them, albeit trivial things ever further down our list of priorities, like designing cute pet costumes. And there will be people newly available to do them at an affordable price, since the robots will have taken all our old jobs and we won't have anything better to do.

II. What Kind of Jobs?

Economic analysis reminds us that so long as human wants are endless, people can be paid to meet them. It doesn’t tell us if this will be a good thing. It doesn’t tell us whether the work will be worth doing.

The history of automation suggests two trends.

First, the improvement of working conditions. Work became physically easier as we moved from hoeing fields by hand to standing over looms and assembly lines to sitting in swivel chairs in cubicles moving electronic paper around and having meetings in air-conditioned rooms. Which makes sense. The tasks that were most physically demanding and unpleasant were the first to get automated (or machine assisted). Yet one result of removing ourselves from doing dirty heavy work to managing (the people who manage) the machines that do the real work is that 'work' itself feels increasingly generic. Whether what you are being paid to do is ultimately supposed to cure AIDS or to sell more photocopier lease-service agreements, from inside it looks and feels very similar. Paper is pushed around and taken home for the weekend; there are deadlines to stress about; many unnecessary meetings are held; there are arguments about the state of the office kitchen.

Second, a reduction in the meaningfulness of the jobs themselves. As we move ever further down society’s to do list - from cat cafes to transformational festivals to Instagram museums to trading World of Warcraft tokens - it becomes more questionable whether what we do makes any real contribution other than to our pay check. The underlying problem is that economic value is not the same thing as human value. As we move down the economic value chain the things we get paid to do are increasingly likely to be valuable only because of the social institutions we happen to have, not because of the kind of beings we are. For example, traditions like Easter generate demand for chocolate candy; mass incarceration policies produce demand for guard labour; market failures like America's healthcare industry produce demand for millions of administrators. By definition our societies are rich enough to afford all this. But it hardly seems the best use of our automation dividend.

If work continues to decline in both onerousness and meaningfulness it will increasingly resemble a multiplayer role-playing game, more virtual than real. Apparently people can be motivated to work many hours a week by this method of collective competitive make believe. Consider the development of the prepackaged chilled sandwich industry in Britain. More than a few people spent the bulk of their working lives and intellectual energies on challenges like developing kinds of bread that don’t get soggy next to a slice of tomato. They even made themselves a Sandwich Association that gives out annual design prizes.

Nevertheless, however economically plausible the gamification of work may be, it is far from good enough for human beings. David Graeber's scathing analysis of the psychological harm of Bullshit Jobs (see also his recent book) lays bare the spiritual and psychological harm of spending our lives performing pointless tasks: jobs that it doesn’t really matter whether or not we do them. Human beings need to do things that matter. Merely having pleasant working conditions and a good salary are not enough.

Unfortunately, from an economic perspective the only permitted measure of value is demand. If we want the jobs of the future to be worth doing - both in dignity and purpose - then we need to realign the economic with human value. That means taking responsibility for the priorities our social institutions create rather than treating them as something that just happens to us. That is a practical, political project.

III Winning the Future of Work

Not all modern jobs are bullshit, of course. Lots of people do get to do things that actually matter - like create beautiful things, or help others, or contribute to humanity’s self-understanding. I am one of those lucky ones. I get paid to solve intellectual puzzles and teach students how to think better. As our society gets even richer thanks to the robots, we can afford more jobs like that. But whether we spend our automation dividend that way depends on the choices we make, not economic principles.

How should a society make that choice? Should philosophers like me come up with a list of what’s worth doing and what is not? No one listens to the edicts of self-appointed philosopher kings – nor should they. Should we organise a big confab and try to all agree on what our society is for, and hence what counts as a valuable social contribution? It seems unlikely that such a conversation would get very far without turning oppressive: why should society get to decide what makes my life meaningful? (Well, perhaps we could all agree on some things, like the betterness of a world without telemarketers.)

My suggestion is that we focus on helping individuals to make their own free decisions about what tasks are worth doing. The laws and mechanisms identified by economics don’t tell us exactly what our robot future holds, but they do help us map possible problems and opportunities. We shouldn’t seek to stop the revolution from coming. But we should seek to shape it so that we can all benefit from the astonishing wealth the robots will bring.

People take jobs because we need them. Wage labour is the main mechanism by which we earn a share of the abundance produced by a market economy and technological advances. Whether those jobs are themselves meaningful is a matter of luck. Billions of robots designed to make our work easier will appear over the next decades. It would be a cruel absurdity to require humans to continue spending the better part of the prime of our lives doing things that don’t matter whether they get done or not. In a society as wealthy as ours will soon be, work should be a choice.

If work is to be a choice then people will need an alternative way to get a share of the automation dividend. A guaranteed universal basic income is the most promising. It is basically a government pension for everyone, not just the over 65s. Taxes on capital would fund a dividend to every citizen that provides them with enough purchasing power for a decent though not luxurious standard of living. If people want more income than that they can still take a paying job as well. But they won’t have to take a job they don’t like because unemployment won’t be a curse anymore.

A basic income frees people to decide for themselves whether a job is worth doing because it increases their freedom to reject jobs they don’t find meaningful, whether because the work itself is pointless or because the conditions are demeaning. This has several consequences for the shape of the future robot economy, especially if it begins early.

With a better fall-back position, human workers with generic skills – those whose labour most closely resembles a commodity, to be bid down to the lowest possible price - will be able to bargain for a better deal. Employers will compete to provide more attractive conditions, including intangibles that middle-class professionals already enjoy, like flexible schedules and respect and consultation in management decisions.

Automation grants our society a new abundance of time. That wealth should be shared among all of us. A basic income does this by granting everyone the right to decide what to do with our own time. We will only accept paid work if what we will be doing with others and for others is worth giving up (part of) the freedom we already have to pursue our personal projects, such as being a great parent or reading everything in the Western canon. From this perspective, time becomes something one spends, not just something one sells. Jobs which call out to us, like environmental educationalist, social worker or philosophical conversationalist, will surely thrive, but they will be structured around our personal lives and needs rather than vice versa. 9-5 jobs that feel pointless will not have many takers.

Wages are driven by demand and supply. If there are lots of people who can do a job, and if they don’t have very good alternatives, then the equilibrium wage will be low. When everyone is already guaranteed a decent income, wages will be driven much more by how people feel about that kind of work. Artists and marine biologists will be content to work for very little, and more people will be able to afford to take such jobs. Tedious, unpleasant, or pointless jobs like cleaning and paper pushing will need to be paid much more to attract anyone. The result is a clear economic incentive for those developing robots: unpleasant and pointless jobs will be the most expensive to use humans for and therefore the priority for routinising and turning over to algorithms. The jobs people find most meaningful will be the ones left for humans to do.


It is presently fashionable to take a fatalistic view of technology. As individuals we are predictably seduced by the superficial attractions of each new shiny invention that comes along - from cars to deep-fried potatoes to social media. And why not? Even if the invention injures society as a whole that is not in our individual power as consumers to change.

If we keep thinking like this, as consumers, the robot revolution will play out upon us in the same way. But that is not our only option. We can reflect on our ethical values, and we can take political action to achieve a world fit for them. The rise of the robots gives us the opportunity to transform work from a tedious necessity that steals away our time and dignity into a freedom that enhances a fully human life.

Note: An earlier version was published in the Institute of Arts and Ideas series: Philosophy For Our Times.