Tuesday, December 03, 2013

Interview with Jim Pethokoukis on robots and jobs

I was recently interviewed by Jim Pethokoukis of the American Enterprise Institute for his Ricochet podcast. The subject was the "rise of the robots" - i.e., what happens if and when technology starts replacing human labor entirely. A transcript, and the podcast itself (if you like hearing my voice), can be found here. The upshot of what I said:

* Technology wasn't as big a cause of declining labor share during the past couple decades as people think Globalization was a bigger culprit. But technology is a much bigger worry going forward.

* Teaching people to be entrepreneurs is one way to stop the rise of the robots from causing horrendous inequality; each person or small team of people can command their own little "army of robots", in the form of a company that they own.

* We will also probably need to resort to some mass redistribution of income. The best way to do this is to have the government subsidize wages, i.e. to pay companies to hire humans and pay the humans more. This is like putting our thumb on the scale against the robots.

Anyway, read or listen to the whole thing here!


  1. Paying people to favor human labor over robot labor? Sounds like make-work to me. Guaranteed Minimum Income seems more workable and humane to me.

    1. Like I mention below, it has the political disadvantages of trying to pass a Basic Income in the US (i.e. it's going to show up as higher government spending and possibly more taxes), without the benefits you'd get just from doing a straight-up Basic Income Stipend.

  2. It ultimately depends on how "intelligent" we can make machines. With some redistribution (like a basic income), you could eventually evolve into a set-up where we're all basically supported on a system that's automated end-to-end for most goods and services, and which provides everyone with a basic minimum standard of living that would still be quite high by modern standards (hopefully). See the "Culture" in Iain M. Banks' Culture novels. If we have artificial intelligence that allows us to do that, which is a big if considering the history of failed dreams and dashed promises in AI research.

    The problem with the "small business" set-up is that most small businesses generally tend to have super-low margins and fail frequently. They're not going to be built on the backs of robots unless robots are super-cheap, and even then consolidation and economies-of-scale are going to hit most of them hard, like what we've seen in the US. There are some sectors where the economies-of-scale favor small businesses, but not most of them - you tend to see large numbers of small businesses surviving in the modern rich countries in areas where there are legal, political, and financial/financing issues with consolidation. Think Japan's hordes of small businesses and small farmers because they blunted the formation of larger retail firms for so long, or small businesses in areas where the city council has kept out Walmart.

    We will also probably need to resort to some mass redistribution of income. The best way to do this is to have the government subsidize wages, i.e. to pay companies to hire humans and pay the humans more. This is like putting our thumb on the scale against the robots.

    It also has all of the political problems of trying to get a Basic Income passed in the US compared to raising the minimum wage, without the benefits and simplicity that you would get with just going the Basic Income route. Those wage subsidies are going to show up on the federal/state/local budget line as stuff that has to be paid for with higher taxes, and higher taxes are very unpopular - even tax credits like the EITC have come under attack from Republicans.

    Whereas the higher minimum wage, regardless of its disputed economic effects, hits the sweet spot in American politics of being Something Paid For By Someone Else Other Than Me The Taxpayer (in the public view), while also Rewarding Work.

  3. I never understand this idea of people starting their own businesses as a cure. Since the VAST majority of new businesses fail what you're saying amounts to "we need to have a massive surge in the number of bankruptcies in order to reduce income inequality" - this strategy doesn't make too much sense to me.
    This is nothing but this common "if you build it they will come" fallacy I constantly here about starting your own business.

  4. I'd like to quickly make two points before I'm forced off to bed. Then later I hope to say more.

    1) I really study and think about this a lot (new book forthcoming by Brynjolfsson and McAfee, and four times as long) with a career in personal finance. How can young people protect themselves against this great risk (and, of course, there's amazing potential for long run good). My current thinking is that you want to skill and educate yourself to be a needed part of a team of humans and robots, for example, a doctor who coordinates, double checks with her unique high level thinking, and helps a team of Watson computers, nurse practitioners, nurses, robots, and technicians.

    This is similar to your entrepreneurial idea, but you can still work for a company, as most people don't have the temperament or desire to be an entrepreneur, with the risks, instability, unpredictability, and responsibilities. This largely came out of what I think was a very fruitful conversation with Nick Rowe, including some nice intuition short models. It starts about here:


    2) Your suggestion to invest in the stock market to own some of your own robots I agree with, and thought of it myself. But an additional method is perhaps to buy up needed raw materials. If robots get really cheap to produce, and you have robots building robots, then if you own raw materials, it essentially may not be hard to sell some to get a robot or two to turn the remaining raw materials into valuable goods. But what raw materials will really be the bottle necks, the ones in demand, for robots to make food, shelter, and so on? In any case, consider that in my home state of Arizona, raw land is ultra cheap per acre. And the powerhouse AZ sun gives you all the power you need to set robots loose on it (Interestingly, my wife always wants us to buy land in Alaska in case global warming fries Arizona and much of the US.)

    1. To clarify (1), a lot of people say, get (high level, abstract, flexible intelligence) skills that a robot/computer can't do. This is largely true, but it implies you do work *instead* of robots/AI computers. I think it may be better to think you try to get skilled/educated to work *with* a *team* of robots, AI computers, and humans, with you as a key member of the team with the high level human intelligence, double checking where the robots/AI computers are weak, and doing the high level abstract analyzing, planning, monitoring, and supervising.

  5. I don't necessarily agree with the premise of wage subsidies because of changes in the structure of labour. There is an argument for it in the short term but even that to me is a limited argument. I think the best solution is to make people be the ones that make the robots like you said and that can be pushed at the end of the human development supply line, education.

    I'll have to listen to the podcast to see if you added more to the redistribution thing.

  6. Technology has been feared throughout history, and apparently today is no different. From the late 19th century Ned Ludd supposedly busting up stocking frames because they reduced the number of workers to make stockings, to the fears of mass production after WWI displacing workers. None of these fears have been realized of course. So, the entire premise of these arguments is false and misleading.

    All of these technologies, including robots, are merely improvements in productivity. Since when is increased productivity bad? What increased productivity does is free people from the most mundane tasks to do more creative and interesting things. Robots will only further this cause.

    The main cause of job loss today is misguided trade policies. In particular, the loss of manufacturing jobs is due to the massive trade deficits that have been allowed by policy for the last several decades with most of that in this last decade and continuing today. The Germans, who have been taking a lot of heat lately, don't have this problem because they manage their trade properly.

    A combination of balanced trade, financial regulation (i.e. Glass-Steagall), and government spurred growth will cause low unemployment and high wages for all. That was US policy from the beginning. Unfortunately, these pro-sovereignty, pro-high-wage, pro-growth policies were abandoned starting in the early 1970's and continue to this day. The solutions are obvious, the political will, nowhere to be found.

    See more at: http://canonicalthoughts.blogspot.com/

  7. Sometimes I have to step back from the world of economics to get a clear view of what is actually desirable. Because actually, having robots do the work for us is exactly what technological progress is meant to do. Utopia is where we all can live without having to work, except if we want to. So yes, more robots please, and at the same time, give people the means to enjoy their leisure time. In the ideal world, time is the constraint, not money.

    1. Thanks for this. I tend to think along similar lines.

      Come on, people. If the technology pans out, we might be able to (for the first time in human history) create a society where nobody has to engage in boring, repetitive work. Do they realize how incredibly valuable that is, how much people from earlier eras of 12-hour days in the fields and factories would kill for that?

  8. Its possible that in the near future, energy will become so expensive that human labor will be economical again.
    Also I agree with J. G. Bullard, if technology free us from the burden of maintaining civilization, we will revert to barbarism.

  9. Honestly, the entrepreneur thing, I think for most people, is bad, or really bad.

    As a successful entrepreneur, I can tell you, it's risky, unstructured, and unstable. A lot of people love routine and predictability, and hate not having it. They want a set schedule and lots of predictability, which usually you don't get as an entrepreneur. And they want a predictable and safe paycheck, and they don't want to have to sell, which you usually have to do with the services of your company. You can hire a salesperson, but the CEO still often has to sell. I could go on and on. It did fit me very well, though; I hate being constrained by someone else's schedule and deadlines, and like (smart and good) risk and adventure. I also like making the important decisions, and so on.

    But another big thing is, if everyone has their own entrepreneurial company with computers and robots, everyone is atomized and alone at work – very very bad for happiness, especially with women. A lot of nerds like being alone at their computer, but all day alone, with no social system, no world of people around you, just terrible for mental health and happiness for most people. Having a lot of co-workers gives you a whole social world, and you don't get that if everyone is alone in their own entrepreneurial venture with the bots. And, even if you have one, two, three co-workers, that's not as rich a social world as a whole office or department of 30, 40, 100+. And the more people there are that you're with day-to-day, the higher the odds you'll find people you really click with.

    Anyway, I don't think we have to all be entrepreneurs. As long as you can work well as an integral and important team member with AI computers, robots, and other humans, then it can be in a company of 30 – 3 million workers.

    It was hard to find the kind of research I was looking for on how many people were right for entrepreneuring. But this is a good article from a University of Texas psychologist:


    1. Bill Ellis7:38 PM

      I think you are right. 70% of rebates are never collected. Sure collecting rebates can be a pain in the rear, but it seems like commanding your own robot army might be somewhat more difficult.

  10. Nathanael11:51 PM

    Why is mass redistribution of income referred to as a "resort", rather than the natural and obvious policy?

    It was explicit government policy in the 1950s -- for almost every government!

    It should be only natural that it's necessary. It's always necessary.

  11. you sound like jonah hill