# Large-Scale Loss Of Jobs To Robots In The Next Few Years. How serious to take this? **URL:** \ **Category:** In My Humble Opinion\ **Created:** [March 21, 2016, 11:38pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808 "2016-03-21T23:38:25Z")\ **Posts on this page:** 11\ **Page:** 6
**Author:** ![RickJay](https://avatars.discourse-cdn.com/v4/letter/r/bb73d2/32.png) [@RickJay](https://boards.straightdope.com/u/RickJay)\ **Post date:** [April 1, 2016, 7:36pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/101 "2016-04-01T19:36:25Z")
> [@SpyOne](#): > > I was just reading an article in Wired from January, and encountered this: > "A 2015 study of Lisbon, Portugal, by the Organisation for Economic Cooperation and Development found that a fleet of just 26,000 TaxiBots—hypothetical on-demand, autonomous carpool vehicles—could replace every one of the city’s 203,000 cars. " I am wholly confident that you could construct a theoretical model of transportation choices whereby 26,000 robot cars could replace 203,000 human-driven ones. I do not for an instant believe that that model would reflect what would actually happen in real life. > [@](#): > > And while I’m discussing cascade: how will a significant reduction in private ownership of vehicle affect the industry that services customizing those vehicles. I don’t mean just the ships that make stretch limos and conversion vans, I mean the factories that make fancy rims. Those jobs depend on people owning their cars. Think of all the jobs lost because people don’t rise horses so much anymore. Of course jobs will gradually be lost. So it has always been. A superior mode of transportation, however, would also reduce the COST of transportation, which frees up resources for use in other areas. Nobody ever looks at that side of it; it’s great that people have jobs making car parts, but the other side of that is that my family has to spend a significant percentage of our take home pay on owning and operating cars, because that’s our only viable option to get around. If we had no need for one or both cars, then it is true people who make parts for the cars we would otherwise have owned will be out of jobs. But the money we spent on cars will now instead be spend, say, on improving our home (jobs for contractors, architects, manufacturers of various things) or furniture (jobs for furniture makers) or vacations (jobs for airline pilots) or savings (less burden on social services when we’re old, taxpayers save money) or whatever. Someone else will have more demand for their job. Complaining about a better way of doing thing costing jobs is, as the old saying goes, like claiming you can improve the economy by running around town throwing rocks through windows so there’s more work for glaziers. Of course, that is little comfort to Bill, who actually will lose his job at the wheel rim plant and is not easily moved into a different job. But that’s why a civilized country invests money in social safety nets. ---
**Author:** ![Peanut\_Gallery](https://avatars.discourse-cdn.com/v4/letter/p/838e76/32.png) [@Peanut\_Gallery](https://boards.straightdope.com/u/Peanut_Gallery)\ **Post date:** [April 2, 2016, 2:47am UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/102 "2016-04-02T02:47:47Z")
I don’t see the issue so much as WHETHER mass unemployment will happen, but WHEN, and how well will people adapt. To the last point, I think people will adapt very poorly, especially in America. I think those who insist that new jobs will come along to replace those lost, “just like for all of history ever!” are failing to appreciate that that trend could change. I don’t dispute that history has shown such a pattern. I am in the “this time will be different” camp. It isn’t just about one technology replacing another, and the economy adapting. It is about the increasing rate of change. I think the Singularity theory is a bit overboard, but that basic trend is true. Things are getting more and more advanced, faster and faster. Tech doesn’t have to reach crazy sci-fi levels, but [del]if[/del] when it gets going fast enough, more than just the buggy whip makers are going to be left without an answer, and in relatively short time compared to the prior paradigm shifts. Our population is just too slow, divided, and intransigent to keep up without some major turbulence. Combine the unemployment problem with the wealth inequality, political polarization, and American dogma regarding welfare “takers”, and I see shit seriously hitting the fan. I don’t see a trend of our society getting ahead of major issues like Minimum Income. More often I see it reacting inadequately and too late (ie Healthcare Reform, Climate Change, Banking Reform, etc). That much all seems clear to me. The only unclear part is WHEN. 5 years is way too short. Maybe 5-10 years to see the writing on the wall. 20-30 years until the real pain hits and people start talking about doing something about it. ---
**Author:** ![Deeg](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/deeg/32/2955_2.png) [@Deeg](https://boards.straightdope.com/u/Deeg)\ **Post date:** [April 3, 2016, 5:12am UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/103 "2016-04-03T05:12:43Z")
> [@SpyOne](#): > > I don’t think that those jobs are 100% going away, but if just 10% of the population decides they can use Uber instead of having a car, that is going to send ripples throughout the economy. Sure, but if you make transportation cheaper then people can go out to eat more often, leading to a rise in restaurants. Delivery services will be cheaper so products that were once too expensive to ship now become commonplace. Everybody can easily see the jobs that are lost. Few can see the new jobs that are created but they always are. ---
**Author:** ![Declan](https://avatars.discourse-cdn.com/v4/letter/d/ecae2f/32.png) [@Declan](https://boards.straightdope.com/u/Declan)\ **Post date:** [April 3, 2016, 8:51pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/104 "2016-04-03T20:51:03Z")
> [@Bosda\_Di\_Chi\_of\_Tricor](#): > > Are we looking at a massive economic collapse, due to job losses? You only really have to look at the reasoning behind the TARPS program, and that had nothing to do with automation. Some people say that we came that close, that the economic map had here be dragons at the edges. We don’t have secondary and teritiary economies, the tax system is too efficient. Automation might collapse the economy later, but we do have other problems now, that could topple the local economy. > [@Bosda\_Di\_Chi\_of\_Tricor](#): > > What degree of human suffering & misery are we looking at? Depends on the govt’s response, as your mainly talking about a self inflicted economic holocaust, and how fast something rises from the ashes of the old economy. TPP and other agreements go right out the window, and tarriffs go up, with the expected rise of prices for what is coming in. Housing, insurance and Credit , gone. For what its worth,I would expect mortgages and car loans to be frozen or written off by govt decree. Unsecured loans, gone, they cant get blood from a stone. What would happen with property taxes, drivers license renewal, plate renewal and all the money that local and state govts take now, is up for guesses. I’d have to look up the great depression to see how they coped that way. Declan ---
**Author:** ![JohnT](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/johnt/32/15048_2.png) [@JohnT](https://boards.straightdope.com/u/JohnT)\ **Post date:** [April 3, 2016, 10:32pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/105 "2016-04-03T22:32:17Z")
> [@Stranger\_On\_A\_Train](#): > > This is a patently ridiculous claim. The notion that technology and engineering fields are easily linearized into the kinds of straightforward tasks that current synthetic intelligence can perform is at odds with the skills and experience it actually takes to perform in these kinds of roles. “Computer programming” for instance, isn’t just about writing code; in order to develop complete, workable, marketable application, far more effort has to be put into developing an architecture, developing an interface to the user, evaluating optimal workflow, and perhaps most importantly, distinguishing the product from other competitors than in actually grinding out and debugging code; in fact, given current modular frameworks and object-oriented design, unless you are doing something extremely novel or technical, developing modern applications are largely a matter of picking the appropriate components and writing some minor glue code to string them together, and then testing the hell out of them to make sure nothing breaks. All of this is creative work that even sophisticated heuristic problem-solving systems are not going to be able to perform in the foreseeable future because it is just too abstract to tell a system what you want and what the end product should look like. Increasingly sophisticated intelligent systems may be used to help write code and certainly to test it in order to make it more robust, but the actual business of developing and supporting applications is going to be beyond the capability of machine intelligence for the foreseeable future. > > I’m not speaking in an abstract sense here; I do engineering analysis and design, and I’ve been hearing for going on two decades now how automated engineering systems and sophisticated CAD tools are going to supplant and ultimately replace the analyst and designer so that some marketing hump can say, “I want a widget that does such-and-such” and have a complete design poop out of the end of some kind of automagical factory ready to go to market. Now, it is true that these CAD systems have practically eliminated an entire class of employees (drafters, checkers, and the librarians who maintain drawing archives), just as word processors essentially eliminated secretaries, stenographers, and typists, leaving just a handful of jobs as executive assistants (with the upside that women in the workplace are no longer assumed to be secretaries by default). However, while the technology has reduced the tedious work of redrawing often poor-quality hand drawings, it hasn’t pared down engineering departments to just a few senior engineers. The analysis tools, far from “automating” analysis and getting rid of expertise, require even more experience and skill to use correctly, as well as requiring analysts to better understand how other disciplines use their results in order to communicate their answers. Integrated multi-disciplinary analysis, where one type of analysis (such as structural stiffness) feeds into another type (such as aeroelasticity), requires considerable expertise in order to establish a problem parametrically in such a way that the interfaces and boundary conditions are correctly established, which is not something automated systems do well, or indeed, hardly at all. And this kind of analysis is increasing required for tightly integrated, highly optimized products that are expected to be robust from first prototype onward. It is not the case that we need _fewer_ engineers; it is the case that we need _better_ engineers and better STEM training that focuses not just on engineering fundamentals but developing practical experience in learning how to utilize such tools to help in critical analysis and decision making. > > The kinds of jobs that are suitable for replacement by robots and/or machine intelligence systems fall into two general categories: those that are menial and can be broken into discrete tasks with well-defined limits of decision-making (e.g. driving a car, cashiering, warehouse stocking, janitorial, et cetera), and those that provide “expert” assistance (i.e. processing data) to actual decision-makers such as physicians, nurses, fiscal analysts, engineers, et cetera, to reduce workload and make those individuals more productive by highlighting important results. The business of actually making critical decisions and doing other creative work will remain the provence of human employees for the foreseeable future. And even some of that “expert” work is still best done by people; despite our best efforts, machine intelligence still cannot read facial expressions or understand subtle cues in spoken language beyond a certain threshold of complexity, and there may ultimately be a limit to just how sophisticated such systems can be in interacting with customers or recognizing emotional responses without some revolutionary conceptual breakthrough in machine intelligence. > > As for the loss of jobs, this will certainly happen, and especially in manufacturing and other menial positions. I’ll be surprised in twenty years if there is a human cashier at a convenience store or a fast food restaurant. (High end restaurants will still hire human waitstaff and cooks because of aesthetics and cache, and because the cost will be borne by the customer.) On the other hand, entirely new avenues of employment may open up, doing the sort of things that machine intelligence is just not good at doing. The gamification of complex tasks such as protein folding, which needs the kind of creative synthesis skills that humans are still much better at than machine intelligence but requires less individual intellect than a mass of generalized cognition, is an area that is ripe to use current workers. The emergence of the “professional blogger”, a vocation that literally came out of nowhere, is an example of how even the technology that automates work can give rise to new sectors of employment. And as the populations of virtually all industrial nations age and start to retire, leaving fewer younger workers, the need for automation and reduction of the more tedious tasks to reduce workload will be critical for a smaller workforce to be adequately productive. Whereas in the pre-industrial era a majority of the potential workforce had to be productive just to achieve sustenance, we currently only have a small fraction of workers doing the basic work to provide food and necessities for an ever-growing population, and the trend for greater productivity out of a smaller segment of the population is crucial to allowing for the kind of “leisure” that permits people to focus careers on scientific advances, commercializing technology, and making sophisticated entertainments. > > The biggest danger of the rise of sophisticated robots and machine intelligence isn’t that they’ll take all of our jobs and leave us unemployed, or that they’ll gain unintended sentience and become our overlords, or other such horror scenarios; it is that we’ll cede such authority and even basic skills that we will no longer be able to fend for ourselves in their absence. But then, that has been true since the beginning of industrialization; hardly anyone living in a modern industrial nation is in any way equipped to sustain themselves in absence of all but the most basic technologies. > > Stranger Aww, it’s been a while since I’ve inspired a wall o’ text, I appreciate it Stranger. 🅱 I hope you’re right, I really do. Maybe robotization (sp) and automation are canards sold to gullible capitalists, the idea that you can wipe out 8,000 jobs with 1,000 machines… and them not taking into account the fact that people have to attend the machines, that customers still want to deal with people, that the increased efficiency in one department causes bottlenecks in other departments… bottlenecks that can only be relieved by hiring more people. That model has worked before and worked well, and there’s no reason it _can’t_ continue. Fortunately enough for this debate, just 5 days ago (on my birthday nonetheless), the good people at [VisualCapitalist.com](http://VisualCapitalist.com) threw out a [nifty little chart of the automation potential of US jobs using _current_ technologies, if these technologies were fully utilized.](http://www.visualcapitalist.com/charting-automation-potential-of-u-s-jobs/) And what they found is that 45% of current US job tasks could be fully automated with current technologies and another 13% of tasks could be automated if the computers could just understand English. When 58% of what you’re paying a staff to do can be done for a fraction of the cost, you’re going to start looking at how to rearrange job functions among the most-skilled staff while [automating away from these assholes](https://books.google.com/books?id=_x8Uj5WLzqEC&pg=PA33&lpg=PA33&dq=automate+away+from+the+assholes&source=bl&ots=bOEG5IUy1U&sig=E-oO40S4Sx5AyOgdY8pkriMEbbE&hl=en&sa=X&ved=0ahUKEwjn8fiKvPPLAhUH1BoKHR9SCXkQ6AEIIzAB#v=onepage&q=automate%20away%20from%20the%20assholes&f=false). And this revolution is not about bringing new products to the people (which is what autos were) thereby replacing old jobs with new jobs building the new thing, this revolution is about how to make what we currently do more efficient, without the pesky effort of having to actually create new products. As far as STEM is concerned, again I hope you’re right and maybe it’s the idiot in me who doesn’t understand the science well enough to figure out why [this can’t be automated](https://www.youtube.com/watch?v=tao0Aes5bko)… but the businessman inside of me is asking “Why am I paying that guy $50k+/year to swish stuff in a beaker?” (at :53) And if the technology is getting better, and the number of job functions to get automated continue to increase, I’m not too sure why STEM jobs get such a pass… and the above-linked McKinsey/VisualCapitalist report has plenty of STEM jobs on it. So, again, the issue facing my 14yo daughter is not one of what is going to happen by 2030 because, shit, she’ll just be a few years out of college by then, but what is going to happen in 2040 and 2050? And while her [Iowa Test Scores](https://drive.google.com/file/d/0B956CzjkwM9vLUFRRmVFQU5ZS28/view?usp=sharing) over the years have shown that she has enough academic and intellectual heft to do well in life, the question still remained - what can she learn now that will be very difficult to program for in the future\*? And the answer is leadership. The ability to read people, capture subtle emotional cues, speak and listen, persuade and motivate. And we work on this, and she’s good at it. She speaks out in class and isn’t intimidated by the boys, leads a number of groups, and (until she became a teenager) was very self-motivated. It will be difficult to automate a job away where the job task is to get people to agree, to work together… but it can be done if all those tasks are automated, leaving Sophia with nobody to lead. But if society comes to that point, well, I have no ideas as to what to do. Lastly, I disagree about the notion that the job of the “professional blogger” came out of nowhere. There were plenty of journalism-degree earning kids in the 1990s who had no choice but to do something internet-wise as the newspapers weren’t hiring anymore - Bill Simmons was one of them and he talks about this frequently - but these weren’t _new_ jobs so much as they were _transplanted_ jobs. And it’s not as if [blogging “salaries” have even come close to replacing lost newspaper salaries.](https://amylynnandrews.com/how-much-do-real-bloggers-actually-make/) Agree to disagree and all that. But, again, I hope I’m wrong and you’re right, but I had to make sure my daughter had an edge and saying “well, we’ll always need computer programmers” just wasn’t doing it for me as an answer to this question 10 years ago. \*I feel, eventually, the answer will be “nothing” or “very little”. But there’s nothing I can do about that nor can I prepare Sophia for such a world. C’est la vie! ---
**Author:** ![JohnT](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/johnt/32/15048_2.png) [@JohnT](https://boards.straightdope.com/u/JohnT)\ **Post date:** [April 3, 2016, 10:43pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/106 "2016-04-03T22:43:40Z")
> [@msmith537](#): > > You only think that because as “sales and marketing of a business consultancy firm” you don’t actually know how to “do” anything. You probably think it’s important too. Don’t get me wrong. I’m in the same field. Nobody knows how to “do” anything. My last client was some financial services company where literally everyone was at least “associate vice president” or higher. They had all these layers of “management” and outside consultants and business analysts and project managers all trying to work out how to solve some esoteric technical problem. So maybe that is the future. We’ll just create a bunch of bullshit jobs for people to feel like they’re doing something productive. > But I kind of feel like we’re a long way off from robots designing robots. Besides, automation doesn’t take away jobs. It changes them. It eliminates job that are tedious, dangerous and repetitive and makes complex jobs that once required highly skilled, educated specialists standardized and automated such that a high school graduate can do them. It lets smart people work on even more complex problems and lets dumb people work on tasks that were once beyond their capabilities. Again, I hope you’re right and I’m wrong and that there will be plenty of jobs for the middlin’ classes, people who want to work doing tasks and not solving problems and who don’t concern themselves with their job when they are off work. ---
**Author:** ![usedtobe](https://avatars.discourse-cdn.com/v4/letter/u/ec9cab/32.png) [@usedtobe](https://boards.straightdope.com/u/usedtobe)\ **Post date:** [April 3, 2016, 11:51pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/107 "2016-04-03T23:51:27Z")
Did you see the clip that Musk released of the Tesla production line? Stamped sheet metal pieces being manipulated in the correct positions relative to each other and then welded together? With exactly ZERO humans involved. Who decided how to make the car - take a model, figure out how to put which pieces together to produce the shell - there are 1000’s of ways to divvy up the thing into small parts. Once you have figured out the pieces, in which order do you assemble them? Finally, which robots are you going to use? Maybe Brand A is a great, general-purpose robot, but Brand B is MUCH better at _these_ jobs - do you mix the make/models of robots (increasing training, replacement part requirements, etc.) or do you eat the inefficiency and use all Brand A? Those decisions are the current jobs of ultra-specialized humans. Where we are now heading, even those decisions will be done by AI. ---
**Author:** ![Stranger\_On\_A\_Train](https://avatars.discourse-cdn.com/v4/letter/s/13edae/32.png) [@Stranger\_On\_A\_Train](https://boards.straightdope.com/u/Stranger_On_A_Train)\ **Post date:** [April 4, 2016, 12:39am UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/108 "2016-04-04T00:39:46Z")
> [@JohnT](#): > > So, again, the issue facing my 14yo daughter is not one of what is going to happen by 2030 because, shit, she’ll just be a few years out of college by then, but what is going to happen in 2040 and 2050? And while her [Iowa Test Scores](https://drive.google.com/file/d/0B956CzjkwM9vLUFRRmVFQU5ZS28/view?usp=sharing) over the years have shown that she has enough academic and intellectual heft to do well in life, the question still remained - what can she learn now that will be very difficult to program for in the future\*? > > And the answer is leadership. The ability to read people, capture subtle emotional cues, speak and listen, persuade and motivate. And we work on this, and she’s good at it. She speaks out in class and isn’t intimidated by the boys, leads a number of groups, and (until she became a teenager) was very self-motivated. > > It will be difficult to automate a job away where the job task is to get people to agree, to work together… but it can be done if all those tasks are automated, leaving Sophia with nobody to lead. But if society comes to that point, well, I have no ideas as to what to do. May the gods save us from the cult of “leadership” and all of the silliness that accompanies it which serves only to benefit Scott Adams and the management self-help industry. Look, being able to make decisions, coordinate work, motivate people to work cooperatively, et cetera are all useful skills, and encouraging your daughter to develop the kind of interpersonal skills that will allow her to succeed both socially and professionally is a laudable effort. But the notion that “leadership” is, by itself, some kind of irreplaceably valuable skill makes about as much sense as baking a loaf of bread out of sand and spit. In order to be an effective leader, you actually need to be good enough at doing what the people you are leading do to understand the challenges they face and how to support and motivate them to overcome those challenges. You don’t need to be the best in whatever field you oversee, but you need to know enough to recognize the difference between good and mediocre work, and more importantly, be able to distinguish a legitimate excuse or complaint from someone just trying to cover up their miserable failure or take credit for other’s work. The absolute worst thing to see in a manager or executive is someone who comes in with a shiny c.v. of “leadership” activities and a gleaming smile, doing out random praise and blithering out pithy sayings repeated from the most recent management technique training or book, and then actually doing fuck all to assist or motivate anyone, and yet, there is an entire industry devoted to cranking out the kind of expert management conslutants that **msmith537** derides, and a culture of worship for these largely valueless “leaders”. As I stated previously, I’ve seen the claims that novel technology using “expert systems” is going to replace actual scientists and engineers for the last twenty-odd years, often by overpriced hired guns coming in and trying to sell their systems and software. I’ve been at two companies that attempted to use such technology to manage “teams” of engineers and programmers around the globe (e.g. in Malaysia, India, the Philippines, et cetera) to reduce their costs. I have yet to see such systems work well, or indeed, even competently, much less actually recoup the often breathtaking costs of implementing them. In one case, a global production management system which was custom-built for the company was in development for five years and deployed over the span of another three years until it was finally recognized for being a total failure and abandoned _in toto_, in large measure because the people architecting and developing the system had almost no real understanding of the needs of the people who had to directly use the system or even a basic notion of the industry it was being used in. We are certainly nowhere near the point of a marketing hack being able to describe a product in general terms and have a automagical factory design, built, test, and manufacture a useable end product. I know that there are people who believe that we could develop this with existing technology, but then, that is because they don’t understand anything about technology. Your link to the video of technicians or scientists processing material (presumably for analysis) makes a point, but not the one that you think it does. While this work is boring and repetitive, seemingly amenable to automation, it actually requires a good deal of skill and decision-making ability to perform correctly and avoid errors, cross contamination of samples, assuring cleanliness of the equipment, et cetera. It is not the monkey-turning-a-crank type of work that you seem to believe it is. Any individual task could likely be automated and performed more quickly and precisely by a robot or purpose-designed equipment, but performing the entire array of tasks that technicians do in varying sequences and quantity makes automation challenging and expensive, hence why we pay highly trained technicians and junior scientists to do the seemingly trivial work of squirting samples into test tubes, a fact you would understand if you’d ever had to work in a lab. In fact, if and when we automate this kind of work it will be a boon because it can free up scientists to do more intellectually challenging work, and will prevent often overworked technicians from making errors due to inattention or lack of oversight, and likely speed up the process and multiply the productivity of tedious lab work by being able to perform tasks round the clock. Note that I’m not opposed to automation, and in fact, I’ve argued here and elsewhere that the foreseeable future of space exploration and exploitation should be largely done with automated systems rather than human astronauts because of the cost and difficulty of mitigating the hazards of the space environment, and the capability that advances in robotics and machine intelligence will offer in the near future is equivalent or even superior to what an astronaut could do when encumbered in a pressure suit and other protective and life support equipment. The same is true for many other hazardous, repetitive, or difficult to physically perform tasks. Recent advances in heuristic methods will likely give rises to changes in transportation that will significantly reduce or eliminate the role of drivers and pilots, although not quite as fast as the most enthusiastic advocates claim, and machines capable of learning by trial and error will open up new areas of tasks which machines can do without having to be explicitly programmed. But human beings are so uniquely good at interpreting multi-valued information and filling in missing data with intuition, and while interpreting human behavior is one of those skills it is far from the only one; in performing novel scientific and technical work, that same core intuitive ability plays a critical role in being able to connect ideas or phenomena together, formulate useful questions or experiments, construct a novel solution, and communicate findings or questions to other people, either in person or by useful documentation. Your belief that work in STEM fields (which encompass a vast array of different types of jobs, tasks, and skills) is just repetitive and easily synthesized tasks belies a complete lack of understanding of what goes on in these fields and how innovations are realized. Automation of the actually simplistic tasks will be welcomed, just as CAD, spreadsheets, and data viz tools replacing hand drawings, calculations, and plots were welcomed by the vast majority of engineers, because it frees them up to spend more time doing novel technical work rather than grinding through calculations or stooped over a drafting table, but it doesn’t mean that engineers and scientists in general will no longer be needed, just the subset of people whose primary focus and training is in doing those repetitive tasks. If there is a danger in the future of employment in STEM fields, it isn’t that trained scientists and engineers will be replace en masse by robots and machine intelligence; it is that technology will allow companies to use skilled labor in countries where education is cheaper and competition such jobs is more fierce. However, the recent experience with trying to offshore technical jobs requiring coordination and decision-making is a cautionary tale with respect to even that kind of job displacement; just having skills doesn’t overcome difficulties in communication and cultural differences. Your daughter will be well prepared for a career in pretty much any field in having good social skills and critical decision making ability, but she will also need to have actual knowledge to perform in that field or even to oversee and coordinate other people in that field, be it law, medicine, engineering, marketing, or whathaveyou. Just having some kind of theoretical decision making template without actual knowledge, or an ability to lead without any notion of where to go or how to get there, is what creates layers of management that do nothing to advance or improve a product, service, or industry. And we don’t even need automation to get rid of those kinds of jobs; we just need to recognize that they add cost to no benefit, or else just accept that we employ people in “a bunch of bullshit jobs for people to feel like they’re doing something productive.” I don’t know about you but I’d rather encourage a child to be proficient at doing something rather than just being some kind of hypothetical “leader”. Stranger ---
**Author:** ![RickJay](https://avatars.discourse-cdn.com/v4/letter/r/bb73d2/32.png) [@RickJay](https://boards.straightdope.com/u/RickJay)\ **Post date:** [April 4, 2016, 12:38pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/109 "2016-04-04T12:38:04Z")
> [@Stranger\_On\_A\_Train](#): > > In one case, a global production management system which was custom-built for the company was in development for five years and deployed over the span of another three years until it was finally recognized for being a total failure and abandoned _in toto_, in large measure because the people architecting and developing the system had almost no real understanding of the needs of the people who had to directly use the system or even a basic notion of the industry it was being used in. This reminds me of the failure of Target Canada, a billions-of-dollars boondoggle that failed in large part because the automated systems meant to control the company’s inventory and distribution failed catastrophically… because they didn’t get enough input from the people who knew what the hell the systems had to do. Every error Target made - and God knows they made a lot of them - was a failure to use the insight of human beings. Aside from the colossal systemic failures, one of the huge mistakes the company made was in replicating its America grocery store model in Canadian stores; the Canadian grocery market, however, is ferociously competitive and its customers spoiled as a result, so Target’s small areas of packaged foods were treated with disdain, drew no customers, and product simply didn’t move. That’s not something a robot would have fixed. ---
**Author:** ![funky\_little\_lee](https://avatars.discourse-cdn.com/v4/letter/f/e95f7d/32.png) [@funky\_little\_lee](https://boards.straightdope.com/u/funky_little_lee)\ **Post date:** [April 4, 2016, 3:18pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/110 "2016-04-04T15:18:55Z")
From a machining perspective: Could robot arms insert blanks into a machine and then remove them when they’re done? That’s already happening. Could a robot actually operate a CNC machine? That would be a lot harder. In addition to being able to quickly move around the area on its own, the robot would have to be able to think and to have “gut feelings” to solve problems. For instance: “There’s chatter on the internal diameter of this part. Is it because the tool insert is broken/worn, or because the specified insert is not in the tool, or because another type of insert would do a better job? Or should I adjust the speeds and feeds? Or is the wrong material in the machine? Or is the coolant flow to the tool not adequate? Or should I change the program and have the tool behave differently? Or do I just have to change the offset of the tool?” Then, whatever the answer, could a robot fix the problem? It would have to be able to loosen a screw, figure out if the insert needed to be changed or just turned around in the holder so a new tip was doing the cutting. It would have to put the insert back in correctly and screw it into the holder again. It would have to be able to find the proper insert in the shop (and ask any machinist how problematic that can be!). It would have to be able to see and measure the part to make sure things are right. It would have to remember how that part ran in previous instances. That’s a lot of back-knowledge for a human, never mind a computer. Possibly robots could replace inserts and do basic measuring of the part, although that would require more than just a simple ability to grasp and let go of things. The decision making process would be extremely hard to program, I think, at least at this point. I think it could come to the point where robotic assistants could help the machinist do some tasks, but only with the supervision of the machinist. So I’m not sure that trying to robot-ize the job would make a lot of sense. ---
**Author:** ![Stranger\_On\_A\_Train](https://avatars.discourse-cdn.com/v4/letter/s/13edae/32.png) [@Stranger\_On\_A\_Train](https://boards.straightdope.com/u/Stranger_On_A_Train)\ **Post date:** [April 4, 2016, 4:09pm UTC](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808/111 "2016-04-04T16:09:30Z")
To be fair, there are advanced CNC machines that are capable of diagnosing many of the normal problems that occur in machining, can track wear and swapout tools, and otherwise deal with many problems, permitting a group of machines running a predefined program to all be overseen by a single operator. This doesn’t necessarily mean that we need fewer machinists; it means, for a given volume of work, a larger number of parts can be produced, and production can run around the clock without dealing with skilled labor shortages or overtime pay. Of course, even the most sophisticated machines can go wrong, and require periodic and unscheduled maintenance, and this usually isn’t frequent enough to justify the cost of designing automation to perform maintenance or sophisticated troubleshooting, so a human operator/supervisor is still required. However, although codes for developing CNC machining programs have gotten far more functional over the past couple of decades, they’re still far from being able to take a random CAD part and turn out a working CNC G-code sequence. Inevitably, they still remove material from an area inaccessible to the tool, or perform an operation that will make it harder to access another feature, or otherwise require fine-tuning before its ready to start production. In fact, automated CNC code generation is one of those examples of how automation and expert systems can aid but not replace a human operator; the code will check toolhead access and recommend the optimal sequence for the fastest operations, but lacks the ability to make judgments outside of its algoritms for sequencing operations. Similarly, additive manufacturing would seem to provide near-magical part creation with minimal human interference, but anyone who has actually had to produce high precision parts with specified mechanical properties using a 3D metallic printer knows of the difficulties and sensitivity to contamination and orientation that the creation of such parts has. There are often features of the design that are necessary to modify or add in order to factilitate machining and printing in order to make the part practical to manufacture, and this often requires some combination of experience and trial & error to work out. Stranger [Previous page](https://boards.straightdope.com/t/large-scale-loss-of-jobs-to-robots-in-the-next-few-years-how-serious-to-take-this/749808.md?page=5)