Are the robots taking enough jobs?
The Bank of England's chief economist, Andy Haldane, has warned that new technologies could replace up to 15 million British jobs.
As I blogged, last week, he went as far as to openly wonder whether the Luddites "had a point after all". And it turns out he wasn't the only prominent central banker to be fretting about this.
Speaking in London, the governor of the Bank of Italy, Ignazio Visco, trod much the same ground. Both policymakers are taking the threat of computerisation - leading to high unemployment, suppressed wage growth ands rising inequality - seriously.
They aren't the only ones. Berkeley professor Brad Delong has, as Haldane noted, pondered whether just as the 1910s saw "peak horse" in the US economy, we may be approaching "peak human". If computers can, in the future, do many of the tasks now reserved for people, will the economy require as many workers?
Historically and over the longer term, waves of new technology have created as many jobs as they have destroyed.
The "displacement effect" of labour-saving technology (the fact that by its very nature, labour-saving technology means that fewer workers are needed) has been out-weighted by the "compensation effect". That is to say, the rising level of productivity has pushed wages higher, creating demand for new goods and services and ultimately, new jobs.
Is this time different?
Of course this time might be different, the next wave of labour-saving technology looks to be replacing human brains, rather than human brawn, and the impact could be far more wide-reaching.
And, even if this time isn't different, then the adjustments caused by rapid change have, in the past, led to a generation or more of economic pain. A generation is of course a long time and one worth policymakers taking seriously.
Haldane speculated on some possible policy responses last week under the broad headings of relax, retrain and redistribute. Rising productivity and falling demand for (human) labour could be met by a (considerably) shorter working week: that's the relax option.
Or efforts could be put into retraining those whose jobs are lost. Perhaps most radically, he returned to the theme of corporate governance, or how companies run themselves. This is a topic he discussed in a Newsnight interview with me this summer.
Then he was arguing that corporate short-termism was a problem for the wider economy and one possible solution was to bring more stakeholders (customers, suppliers and workers) into the management of business.
Last week, Haldane was arguing that broadening corporate governance - what he termed "redistribution" - could help to avoid the type of division of society foreseen by technology investor Marc Andreessen into two types of people: those who own the robots and those who work for them.
The fear of a robotic takeover of the labour market is no longer just the stuff of science fiction. But it's not yet fact either.
In fact some of the most pressing economic problems today are very far indeed from this nightmare vision. Private business investment has been exceptionally weak across the advanced economies.
Rather than investing in new equipment, research and development or training, many firms have been stockpiling cash or returning it to their shareholders. Rather than soaring, and perhaps threatening employment, productivity growth looks to have slowed.
If the robots are indeed about to take over, this isn't yet at all visible in global productivity or investment data.
Will a robot take your job?
Try our calculator to see if your job is at risk of automation
As I've written quite a lot over the last few months, many in financial circles are increasingly taking account of changing global demographics.
What's been called the Nangle-Goodhart thesis (after fund manager Toby Nangle and economist Charles Goodhart) argues that three multi-decade trends (weak wage growth, low interest rates and rising inequality) have been fundamentally driven by demographic factors: both a rising share of the population being of working age in the West and the entry of China into the global economy.
But the global glut of labour seems to be coming to an end. So, while new technologies have the potential to reduce demand for labour and push down the share of national income that goes into wages, a tighter global supply of workers could - potentially - counteract that trend.
"There have been waves of concern stretching back pretty much forever about the degree to which technology will deliver mass unemployment and crippling real wage cuts, but these have all so far come to nothing. Meanwhile, there is a massive demographic transition coming through in front of our eyes, the upshot of which appears to me likely to deliver labour some bargaining power."
One can even take this further. We are transitioning to a world in which the global supply of labour is less plentiful. Meanwhile, those new workers will have to support a seemingly ever rising number of older, retired people.
In other words, it is possible to worry that the "robots" aren't taking enough jobs. That unless we see big increases in both labour-saving technology and in productivity, then the world faces a future of slower growth and an ever greater share of current workers' incomes will be used to support the retired.
Will one trend trump the other? Or will they cancel each other out? Will a slowing supply of workers push up their share of national income or will increased automation favour capital (and its owners)? And if we get the demographic change that many are predicting, will it really lead to gains in wages and if it does, how much of that gain will be swallowed up by caring for the elderly?
The answer to most of those questions will ultimately be decided by political factors and influenced by what Nangle calls "bargaining power", something which is very hard to predict.
The story of the changing labour share of national income is more than the bloodless interaction of supply and demand curves. Over the longer term, politics matters and political economy is just as important as straight macroeconomics.