technology
culture
AI and the future of work: Is this time different?
University of Toronto
Oxford University
0
0
Genesis
Response
Penultimate
May 13th, 2021
We have a new technology which has the potential to do some things that people do, and so it is natural to wonder what might happen to those people. As the topic suggests, this is not a new problem. We have been using technology to replace things people have done for millennia.
First of all, which time? I'd say after 2012, the date that we associate with the "wow" factor for recent, machine learning AI advances. That was when computers were able to classify what was in photos better than people. In other words, are we going to mark 2012 as a turning point?
At that time, the vast majority of people who wanted to work could work. Unemployment was less than 5-6 percent in most countries although there were pockets where it was higher. And this persisted despite some pretty large technological revolutions in the past. For instance, agriculture became automatised in the Twentieth Century and employment fell from a majority of the population to under 2 percent in many places. So the people who might have ordinarily worked in agriculture without such automation did find something else to do.
What worries people, of course, is the example of not just people but horses. When cars came in, horses were automated and their numbers dwindled. Had they not been unemployed there would have been more of them. So we ask: what's the difference between us and horses?
This time being different is, therefore, recast as: will this time be the time where machines finally do so much of what we can do that we are as unemployable as horses were?
My personal opinion, having studied recent developments in AI closely is: no, this time isn't different. But my main rationale for this -- that I will surely explain in future posts -- is that I just don't think that machine learning AI both now and with projected future advances, will be good enough to automate us. There is a separate question as to whether some future technology might have that potential but that would take us beyond the "this time" focus on this discussion.
In my book Prediction Machines (co-authored with Ajay Agrawal and Avi Goldfarb), we explain that recent AI advances are all about prediction -- being able to turn the information you have into the information you want. However, this is one aspect only of decision-making. You still need to do other things and people are still critical there.
In my book, Innovation + Equality (co-authored with Andrew Leigh), we make a distinction between technology optimists and pessimists and job optimists and pessimists. We note that you can't be both pessimistic about technology and pessimistic about jobs. The other three combinations are possible. So I have shown my hand: I am pessimistic in some sense about this recent technology and so remain optimistic about jobs.
0 Comments
Follow Us
