Marc Andreessen, a programmer who made a fortune off one of the first web browsers and then a bigger one by investing the spoils, predicted some years ago that the world would soon be divided into those people who told computers what to do, and those who are told what to do by computers. It is a pithy demonstration of the half truths that drive Silicon Valley. The task for humanists must be to ensure that the other half of the truth is valued and acted on. Computers are of course turning the world upside down, and transforming many of the jobs that they don’t abolish. The process seems absolutely certain to continue: a recent OECD study suggesting that only one in six jobs in industrial countries will disappear as a result of automation has been hailed as a wonderfully optimistic counterblast to earlier predictions that nearly half the jobs in advanced economies would disappear.
One job in six would – will – still involve a major social transformation. Unless regulatory action is taken, this will not be a transformation for the better. For one of the characteristics of this wave of automation is that many of the new jobs which appear are worse and less interesting than the ones they replace, in part because the humans doing them are fitted into the decision-making hierarchy below the people who write the algorithms that control them and the computers that interpret the programmers’ intentions. To that extent Mr Andreessen was right. But the new work has other disadvantages. It is much less secure and less well paid. Those factors are not determined by technology, but are the consequence of deliberate political decisions that technology serves to obfuscate. This is already obvious to young people, and drives a lot of the support both for Jeremy Corbyn and, separately, for populist nationalist politicians in Europe, who represent the threat to security as coming from immigrants more than automation.
The OECD report reaches its relatively optimistic conclusions by analysing job descriptions more carefully than previous estimators did. The kinds of skills which the report sees as hard to replace are those involving social knowledge and interaction. This optimism may be to some extent unwarranted, for two reasons. The first is that computers are getting far better at analysing and classifying both sounds and imagery – the kinds of skills which are easily mistaken for understanding. Software which runs on pretty ordinary computers and can pick the cars and the pedestrians out of a video recording of Indian traffic (in which cars, lorries, people and so on are mixed apparently at random) is already freely available to anyone who wants to play.
The second is that just as computers are being trained to accomplish more, we, the public, are being trained to accomplish – and to expect – less. Increasing amounts of customer service require us to interact directly with computers at the other end, whether by filling out forms or dealing with voice recognition systems. At the same time, those humans still employed for social interactions are increasingly scripted and regimented; so that in dealing with a modern bureaucracy it makes no difference whether there is a human or a robot on the other side of the transaction. No doubt real human servants will increase in value as status symbols for the rich, but this is hardly a future to anticipate with pleasure.
A more human and equal future will require a great deal more than simply training and re-education, necessary though those will be. It will require the kinds of politics and economics in which no one may be thrown on the scrapheap as useless to society.
