The National Academies' Center for Education has been working on a study of the future of computers and work. By 2030, they ask, what kinds of capabilities will computers have; how well will those capabilities prepare them to do jobs currently done by humans; and what proportion of the workforce might be displaced or rendered unemployable?
The results are rather scary. After looking at trends in machine vision, speech, reasoning, and movement, and estimating how important these are for doing various kinds of work, the author estimates that displacement rates could be over 80% in some fields-- sales, administrative support, food preparation, and personal care. These are also the sectors that employ the largest number of people. The safest fields for humans? Law (6%), medicine (10%), science (10%), and engineering (11%)-- fields which currently employ the smallest number of people.
[Source: Stuart W. Elliott, "Projecting the Impact of Computers on Work in 2030," p. 37, available online [PDF].]
A draft of the article is available online, and it has a lengthy description of its methods.
At first glance, it looked to me like there was an obvious flaw in the study. The high rates of replacement in "education, training, and library" suggested a systematic under-valuing of tacit knowledge or the social dimensions of work. If you assume that education is learning facts, and librarianship is finding books-- and nothing else-- then these high displacement rates would make sense, but otherwise they wouldn't. However, the relatively low replacement rates for repairmen and protective services suggests that that's not so. Any method that undervalues teachers isn't likely to also overvalue Larry the Cable Guy.
So the best places for humans in the future will be litigious, technocratic societies that spend a lot on health care. Actually, that sounds a lot like California.
Interesting that legal rates such a low level of displacement. I suspect that the legal industry will see greater displacement than education and librarianship, actually. Although flashy TV-style argumentation gets the attention, much of the work in the law trade is trying to determine whether regulation A is applicable to action B.
Posted by: Jamais Cascio | August 29, 2007 at 05:32 PM
I don't buy the idea that so much of legal practice is immune from automation. Certainly large parts of trial law, and areas where the law is unsettled, are ones where human judgment and craft really matter; but a lot of legal stuff-- most wills, for example-- just requires careful attention... indeed, an almost algorithmic approach to procedure that guarantees nothing is left undone, things are filed with the right offices on the right paper, etc.. If that can't be automated, nothing can.
Posted by: Alex Soojung-Kim Pang | August 30, 2007 at 10:34 AM
My thought in response to this is that job protection for legal services occurs not because it cannot be automated, but because law as an institution, and lawyers as a profession, are more powerful than other professions. And so they are able to make sure that less is automated. Carol Heimer used to call law a "master institution" (or something close), because while it competes with medicine, family, it also - uniquely - gets to define the rules of the game.
So perhaps what we're seeing is not automation in response to the technical qualities of occupations, but the technical qualities of occupations in interaction with professional power or something like it.
Posted by: Peter | September 03, 2007 at 04:56 AM