Just what is it about your job that a machine won’t soon be able to do?
An interesting article by Erik Brynjolfsson and Andrew McAfee, “The jobs that AI can’t replace”, in which the authors of “Race against the machine” and “The second machine age”, explore some of the areas where they say artificial intelligence won’t be replacing humans any time soon: creative tasks, which include scientific research, writing, and setting up businesses; social interaction (running companies, leadership, sales, negotiation, caring for people, nursing); and tasks requiring physical mobility or agility, like gardening, or cleaning. The authors encourage workers and governments to take advantage of the skills required in these sectors in a bid to delay the progressive impact of job replacement by machines.
Hmm. I have my doubts. Firstly, their choice of jobs: back in 1996, when I arrived at UCLA to do my doctorate, I met people working there on developing algorithms able to carry out automated negotiations, the results of which were pretty decent compared to what their human counterparts could do. Needless to say, these are technologies that have come a long way since then, and not just on the basis of Moore’s Law. Financial markets around the world are experimenting with negotiation techniques based on algorithms trained in this area, able to interpret all kinds of signals, and even to develop buying patterns aimed at confusing or provoking reactions from others playing the markets.
As for scientific research, I think it would be reasonably straightforward to develop automatons able to manage the complete bibliography or sources of a given subject, and to understand and indicate which are the most relevant and what relationship exists between them, as well as to create the structure of a review paper that could be used as the basis for anybody who wanted to carry out investigation in a given area. Whether or not doing so is a good idea, given the human learning and research training that takes place during this task of putting together a review paper, which is one of the first things a doctoral student is assigned to do.
As far as social interaction goes, well, we’re seeing that a growing number of bots of different kinds are acquiring the intelligence required to manage a growing number of community management tasks, and not just those aimed at getting men to pay to talk to a woman who in reality is a female bot (Ashley Madison comes to mind…), although they are able to deal with significantly more complex tasks.
When it comes to jobs that require manual dexterity, such as assembling components, we might point to the case of the Chinese factory that assembles electronic components and that has replaced 90 percent of its workforce with robots, improving productivity and quality in the process. And the remaining 10 percent of the workforce? Sorry, but they’re not involved in adding value to the products: they simply look after the robots.
The possibilities for replacing humans with machines depends on a wide range of technologies: machine learning, which is my main area of interest, is now able to carry out a huge number of tasks that we once believed to be the reserve of humans, in the process improving every aspect of the process.
I see no reason at all why the areas Brynjolfsson and McAfee use to illustrate their argument will be safe from machine substitution, and not just in terms of the next decade, but much sooner. A machine is perfectly able to analyze the variables involved in decision making, and do so more efficiently than a human, whether that applies to driving a car or running a particular area of a company.
We may still be in the theory phase, and I would certainly recommend Brynfolfsson and McAfee’s great books. But if there is one thing we can learn from reading them, it’s that we should all be thinking about which aspects of how we earn our living we can protect from the machines. Not panicking, not being apocalyptical, not fearing or trying to run away from them. Simply, accept it as a fact of life, and learning to live in that scenario. Very soon, we will be required to make decisions in this regard that will have a profound impact on our future. For better or worse.
(En español, aquí)
