Andrew McAfee and Erik Brynjolfsson -- Machine platform crowd ================================================================ Some other books in the same topic area: - Erik Brynjolfsson and Andrew McAfee: "The second Machine age" - Jerry Kaplan: "Humans need not apply" - James Barrat: "Our final invention" - Martin Ford: "Rise of the robots" - Andrew McAfee and Erik Brynjolfsson: "Machine platform crowd" - Yuval Harari: "Homo Deus: A Brief History of Tomorrow" And, there is this article: "Robot-proofing your child's future", The New York Times, 12/14/2017, p. D1. Some issues ------------- Some of the issues that are addressed by these books: - Machines will do all the tasks that we are currently employed to do and we will all be out of work. And, the range of tasks and types of work that machine will do and will replace human labor is larger and wider than we expect. - The capabilities that we are giving our machines (or that they are acquiring, in the case of machine learning) will become dangerous; we will lose control of our machines; perhaps, even, our machines may conspire against us and be dishonest with us. - The algorithms and code that machines will use, especially for machine learning, are complex and non-logical. They employ, for example, pattern matching that is "trained" using large volumes of samples. We will not understand or be able to justify the decisions made by these algorithms. - There will be tasks that machines will perform better than we humans do, for example, with more precision, more accurately, with less (human, male, racial, etc) bias, with fewer mistakes, more safely, etc. How will we decide when machines should be allowed to perform those tasks and when not? How will we decide when humans will be allowed to perform those tasks and when not? I can imagine a time, for example, when under some conditions, humans will not be allowed to drive automobiles, because they kill too many people when allowed to do so. Many of the issues that are addressed in "Machine platform crowd" in particular are ones that attempt to provide guidance for managers of on-line companies. Who will perform the task and make the decision -------------------------------------------------- Safety -- There are some jobs that can be performed by either machines or humans. Driving a car will soon be one of these. If we find that machine driven cars are safer, that they kill and maim fewer people and cause less damage than human driven cars, will we continue to allow humans to drive cars? Will we, perhaps, limit when and where humans can drive cars, e.g. only on city streets but not freeways or only within 100 yards of your own driveway etc? And, if we make these kinds of restrictions on human driving, how will we justify them? In what sphere (legal, moral, practical) will make those restrictions? Very few loans are now made based on human evaluations. We know that a human loan evaluator will be biased, will show racial and sexual prejudice, will make mistakes. For a number of reasons, in person interviews have proven result in bad decisions. The changing nature of employment and work -------------------------------------------- There is a good chance that there will be many fewer jobs. There is also a good chance that many of the jobs that we do have will not provide the ability to earn a reasonable standard of living. And, certainly, the distribution of kinds of jobs will change. There will be more jobs that require technical training, skills, and knowledge. There will be lots of jobs such as service sector jobs, even ones that require experience and skills, that do not pay well and do not provide benefits (medical insurance, vacations, sick pay, etc.). It's also likely that many of these jobs will be in the "gig economy": they will not, for one, provide a dependable income stream. Yes, it's a little comforting to believe that these employment problems and disruptions can be solved by more training and education. However, there are lots of people who are going deep into (student) debt to get college educations that often do not get them jobs that enable them to pay off those debts. And, although we are perhaps reassured when we read newspaper stories about younsters getting technical training with computers and electronics and the like that will help them compete and work in the high-tech workforce and economy. But, I am not confident that there will be a large enough number of those hi-tech jobs, nor that they will pay well enough, and still less that most of us will have the abilities need to perform in those jobs however much training we receive. It's very unlikely that there are remedies for these problems and disruptions. And, it's even more unlikely, that our U.S. Federal government will be willing to react to and fix these problems. After all we have lots of members in our national legislature who believe in "cowboy capitalism": the idea that we should all be able to help ourselves, and should not depend on our government to take care of us. McAfee and Brynjolfsson seem to be strong believers that digital platforms will produce lots of new and innovative capabilities (for our computers, our mobile smartphones, and for other devices as well). I believe that, too, but, I'm not so sure that all the results will be so positive. And, what also worries me is that we will be soothed and pacified by these shiny and attractive new toys, and because we are, we'll allow, without push-back, these drastic changes to employment to happen. McAfee and Brynjolfsson also believe that digital platforms (and other changes to our economic landscape, too, I suppose) are going to make lots of goods less expensive. That means that we can buy more of them and we will have more money left in our pockets and bank accounts after we buy them. That's likely true, *but* since they are so much more cheap to produce, we (workers in particular) can earn less by making them and fewer of us can earn even that. This is especially true of goods that are easily and cheaply reproducible, electronic media in particular. As reproduction and transmission/transportation costs fall, who will be able to make a living providing those goods? And, when you factor in the fact that owners and managers, especially top level managers, are taking a larger percentage of the revenue, this trend will bite hard. McAfee and Brynjolfsson's concerns become even more worrisome when you consider that there will be more and more support for those who are attempting to develop AI software. There is open source software available for programming languages like Python. And, I'm already seeing lists of how-to documents and tutorials on the development of AI software and machine learning systems. Some of us will be trying to learn how to develop that software. And, can you imagine what it will be like when the next generation of students graduate from computer science departments when they have been expected to AI and machine learning software for their class projects? That's a very sobering thought. Open source software ---------------------- I'm a software developer; I'm retired; but, still I'm sensitive about issues of employment of software developers. McAfee and Brynjolfsson talk about demand curves and price curves of complementary goods (in chapter 7, "Paying complements, and other smart strategies"). These goods have several characteristics of interest: (1) Because they are software, they can be copied and distributed at almost zero cost. (2) Since they are open source, they conceivably replace software that might have been developed for pay by programmers who might have earned a living by doing so. (3) They increase or facilitate the sales of some device or application which was developed for pay and which does earn money for some organization and the people who are employed by it. One response to this uncomfortable situation is to say: Developing open source software is a choice. If you want money in return for you effort, either do something else, or figure some other way to earn money, e.g. by providing expertise and services for pay, by selling related products and extensions or a "professional" version for money, etc. The dangers in machines/computers that are too smart ------------------------------------------------------- Here we are stepping into areas of science fiction and speculation. We do not really know in what ways computers will become so smart that we will fear them. We do not know what practical ways we will have so that we can control and limit their capabilities. We can, of course, imagine a future in which there are so many computers near us, in our bodies, in our homes, and where their activities are so many and so complex that we will not know how to control them. So, a few points, in an attempt to reassure us: - It may be that for some purposes and uses, we will not want to be able to control or restrict computers around us, because they will to a better job of making decision than we are with our biases and tendencies to make mistakes. - We may see the use of rule-guided limits and restrictions on what computers are allowed to do. This might be analogous to the rule-based software that controls the fire-walls in our wireless routers. There are lots of bad things that computers out on the wider Internet can do to the computers and devices behind my router and inside my LAN (local area network). I don't know what all those bad things are nor do I know how to prevent them. But, I (mostly) trust my router to stop the traffic that would allow those bad things to happen. I can imagine (in a vague way, details to follow) that in the future, I will be protected from the machines around me and inside me by analogous sets of rules enforced by devices whose purpose is to shield me from other devices. This scenario is a stretch, but after all, we *are* talking about an imagined future. In general, I'd say that we are talking about the future. We have a lot to learn in order to be able to deal adequately with the changes in that future and devices and capabilities and threats etc. that those changes bring. We'll make mistakes at first at least. We'll learn and we will gain new capabilities, new devices, and new software that will enable us to deal with those changes. The future sounds interesting and exciting to me. Let's hope I do not rue the time when I looked forward to these changes. .. vim:ft=rst:fo+=a: