One very powerful reason for thinking that the future impact of technology on the demand for labour may be different from the present, is that we have seen changes in the nature of technological progress in the past. For example, the current consensus view among economists (see Goldin and Katz, 1998 or Acemoglu, 2002, for recent summary discussions of this) is that the impact of technology in the early days of the Industrial Revolution was to reduce employment among skilled artisans (and the wage declines that resulted from this can be argued to be the origin of Marxs immiserisation hypothesis, another lesson about the dangers of extrapolation from the past into the future).
Another particularly pertinent example is the decline in the employment of servants in the middle of the 20th century. The jobs done by servants were non-tradeable and dependent on the presence of the rich in exactly the same way that I have argued is increasingly characteristic of the employment of the low-skill workers today. But, the share of workers in domestic service declined rapidly from a peak of 8.2% in England and Wales in 1931 to 1% in 1971 and from 6.5% in the US in 1930 to 1.7% in 1970 (figures from Singelmann, 1978, who also has data from other countries that show a similar trend). The most common explanation for this collapse is the invention of the vacuum cleaner, the washing machine and other domestic appliances i.e. technological change (though declining wage inequality may also have played a role). If that has happened in the past, then it can happen again.
So, let us consider what would happen if machines could be cheaply employed to do mundane human tasks like cleaning the house. This is not beyond the bounds of possibility: in the field of robotics there is a lot of research about designing machines that can do hand-eye coordination and 2003 saw the launch of the RoomBa RoboticFloorVac a moving disk that can be left alone in a room to clean it. The manufacturers web-site is full of praises for it though one must realize it still has considerable limitations as it can only traverse uneven floor transitions up to one half-inch tall (i.e. if you want it to go upstairs, forget it) it cant do thick carpets and it might occasionally throw itself down the stairs, all characteristics one might think of as problematic in a human cleaner. The nature of technological progress is partly determined by what is scientifically possible but also by what is economically worthwhile (see Acemoglu, 2002, for a recent analysis of this). For example the Economist of March 13th 2004 contained an article entitled the gentle rise of the machines about the increasing use of robots and quoted the inventor of the first industrial robot (Unimate, employed by General Motors in 1961), Joe Engelberger as saying that care of the elderly is the opportunity the robotics industry should be pursuing as every highly industrialized nation has a paucity of help for vast, fast-growing ageing populations, something that can be readily understood from the main argument in this paper.
But, while these technological changes might happen this is probably not something that that will happen anytime soon (that has a bad track record - compare the representation of the year 2001 in the film 2001 made in 1968 with the reality) or of anything that will inevitably come to pass. So let us think about it in a what if kind of a way: that is why it might best be described as science fiction economics. Science fiction is generally concerned with transformations of society (otherwise it is just fiction) but the economics in most science fiction is terrible.
It is a common theme in science fiction that machines will come to be intelligent and have a mind of their own and then, through some act of revolution, come to dominate humans a good recent example would be The Matrix. In economic terms one is envisaging in this film a violent change in property rights with a change from humans owning machines to machines owning people but, in that film, humans are then used as a source of energy, suggesting that the machines understanding of the laws of economics was somewhat deficient as humans could not possibly be an efficient source of energy.
But this emphasis in science fiction on change through violence underestimates the potential impact of change through the laws of economics. For example, in science fiction the problems for humans often start when machines get as intelligent as them and have a mind of their own. In contrast, an economic approach would suggest that the problems will be worse when they are not that intelligent and do not have a mind of their own. One of the big problems with employing humans to do jobs is that they do have a mind of their own, so that the extraction of labour from labour power (as Marx put it) or incentivizing workers (as business school professors put it) is a problem. An obedient machine is far more of a threat to most humans than a thinking machine.
In a world where menial jobs can be done more cheaply by machines than humans the demand for the labour of the least-skilled would collapse, quite conceivably below the cost of creating and maintaining a human. Unless they own capital they would be unable to obtain an adequate source of income by selling their factor endowments. We would finally have ended up in the world of Marxs immiserisation hypothesis in which the wages of many workers would decline in absolute terms and the levels of inequality between humans are determined by inequalities in the ownership of capital. In this scenario it is the distribution of capital ownership that is crucial and it is important to know whether this tends to become more or less unequal over time. Recent papers on France (Piketty, 2003) and the US (Piketty and Saez, 2003) suggest that progressive taxation (and, less optimistically, wars) may be able to put a brake on rising inequality in capital income but, left unchecked the distribution of wealth would tend to become more unequal.