Hacker News new | past | comments | ask | show | jobs | submit login
The Last Robot-Proof Job in America? (newyorker.com)
41 points by hhs 79 days ago | hide | past | web | favorite | 65 comments



Robotics still seems a long way off being able to automate many physical jobs - would you trust a robot to look after your aging parents without hurting them?

Likewise jobs like computer programming seem pretty 'robot-proof'.

It seems only repetitive, predictable work can be automated and we shouldn't waste human labour on those tasks in any case.

Personally, I think the threat of automation has been greatly exaggerated in the media, it is true that self-driving cars would have a huge economic effect but progress appears to be slowing down in that area.


would you trust a robot to look after your aging parents without hurting them?

Having seen how little many humans care about other peoples aging parents under their care, I'm not so sure robots would be a huge step down.

edit: but as others have pointed out, robots will/should be there to help humans do the hard parts of their jobs not completely replace them. So in the future we'll probably see a human and a robot doing what used to take 3-5 humans, not simply a fleet of robots replacing all humans.


> Having seen how little many humans care about other peoples aging parents under their care, I'm not so sure robots would be a huge step down.

Which is an issue. It's the old addage: I don't have to outrun the bear, I just have to outrun you.

If the robots are only maginally better than the humans, then they will out compete the human workers and we will have robots that may still be very bad at doing the task; potential failure all the same and with bad patient outcomes.

> So in the future we'll probably see a human and a robot doing what used to take 3-5 humans,

In Chess, these amalgams are called 'centaurs' and they consistently outpreform humans and/or the best computer programs. It's a good idea in chess, as most centaurs are made with very good human players. It's when the humans start to rely on the computers that you get into real trouble.


True, productivity growth is very low compared to the past. It seems that the long recession coincided the hyping of certain new tech (self-driving cars, drone delivery) leading some people to wrongly believe that we were on the verge of robots causing mass unemployment.

On the other hand, I think we might not have really dealt with the changes from mass automation in the past. If you look at things like "bullshit jobs"[1], or large sectors of the economy that provide negative value to society, it seems like previous automation freed up huge amounts of labor that simply haven't been used to improve society. Since that's already happened, though, it's much harder for people to see it as a problem.

[1] https://www.newyorker.com/books/under-review/the-bullshit-jo...


Taking care of parents without hurting them is very easy. The bigger problem is whether they can take care of them at all.

Also the real threat of automation is wealth inequality. If workers own (a share of) the robots that replace their jobs then there is no problem. If all the robots are owned by corporations there is a massive problem.


Actually, wealth inequality is a big threat /to/ automation because lower wages means automation isn't economically efficient.

The first places where automation is put in place is where labor is expensive. That is, both geographically and industry-specifically.

It's not the factory workers so much as the accountants, lawyers, and doctors that are marginalized first. Software is cheaper and the reward greater.

And with respect to geographically, Scandinavia employs the most robotics to augment manually labor jobs. They will continue to lead in this area.


I found that interesting. I've also heard it said that unemployment is higher in France compared to England. But their (France) productivity is higher. I wonder which is better...


Productivity in my opinion - there is no point in employing people if it isn't efficient.

But it seems most societies still seek jobs for jobs sake.


Agreed. We need to migrate to worker owned and run businesses ASAP.

I think there's some pretty significant blowback coming. Poor people don't just disappear.


Work is going to be here for a very, very long time. There are aspects of it that can be automated, like some parts of manufacturing.

I think the bigger problem we are going to encounter is a retiring workforce. And the lack of knowledge transfer in a lot of industries. I recently walked into an office with a PBX and no one knew how it worked. The servers were installed in a basement and the documentation was damp, moldy, and crawling with bugs. What should have been a simple transfer to VoIP now requires tracking down lines and real downtime.


> I think the bigger problem we are going to encounter is a retiring workforce.

Or a workforce that is "replaced" by young, fresh-off-university people that are desperate for a job that pays anything to survive with no benefits, while at the same time the old guards with real, proper incomes are fired as they are vastly more expensive...


Isn't it strange that his has become such a problem, people who can't or won't retire, once pensions were destroyed?


I was talking with our 401k company guy about this. But basically, he told me most 401Ks who were invested before 2009 are just becoming whole now. If you were planning to retire before 2009 and converted to bonds before 2008, like you were supposed to, you were okay. And if you started contributing after 2008/2009 you did well. But for a lot of folks, who contributed on the riskier side of things (basically anyone under 55) you lost 10 years. Anyone that planned to retire at 65-67 is now having to do so at 75, otherwise they will be underinvested at retirement.

I anticipate you are going to see a large wave of workforce retirement because there is such a large demographic who had to delay retirements. And that bill is now coming due.


This is another, separate problem - but yes, real indeed. This kind of stuff happens when government services get defunded or outright destroyed...


It is easy to imagine an elder care robot that is humanoid in features and attempts to directly replicate what a human does. But that is rarely how it plays out. Instead you get technology that solves a single practical problem. For example, a washing machine, ready meal, dishwasher, or hi-tech toilet. All replace activites that require a lot of labour to do by hand. Maybe there are other practical technologies that would be of utility to the elderley and help maintain independance.


Yes. We already have retirement complexes where eg chairs have sensors which alert someone nearby if someone is sat down for too long. Net result, you have less need for someone to 'check in' on people (but then you need to find another way to give people social contact). It's about using human labour more efficiently, not replacing humans.


> would you trust a robot to look after your aging parents without hurting them?

I often see these kinds of examples with 'non-automatable' work. I agree that the full extent of this might not be automatable, but there's probably opportunity for efficiencies through automation, which won't make all people reduntdant, but might decrease the number of people needed to provide this care.

One thing I could think of is camera systems with anomaly detection to watch over people to make sure they're not hurt or something. Or biometric devices that provide feedback to a centralized system. Sure, people might be weary, bot 15 years ago people were stil weary of mobile phones too.


Like computers, robots should be seen as workforce multipliers. A robot that helps lift elderly out of the bed with human supervision makes a two person job a one person job.


The typical way automation takes job, is not by replacing them but by increasing the efficiency of the task being done, therefore needing less workers to do the same job.

>would you trust a robot to look after your aging parents without hurting them?

Physical robots are not there yet, but you can already get an improved efficiency by camera monitoring. Using computer vision and pose tracking you don't have to always be present, you only need to be able to dispatch someone if a problem is detected.


> would you trust a robot to look after your aging parents without hurting them?

My grandparents certainly could use a proper self-driving car if it were able to drive on the Croatian "magistrala" (central road on the coastline, very dangerous as it is), and my parents are really satisfied with the Roomba my sister and I got her for xmas two years ago. Probably what could be very useful to both is a robot that does laundry and dishes, basically the two tasks that are most demanding on their backs.

For people in retirement homes or with disabilities, robotic assistance in basic bodily functions (=showering/keeping clean/eating) or cooking probably would be a godsend.

This is the kind of robotics we need to free up (caretaker) resources, but unfortunately these tasks are simple for humans but exceptionally difficult for robots.


One of the problems with these type of robots, is misuse. People who could become capable of doing for themselves, instead become more and more disabled because the lean on automation and won't work to develop their own ability. We already see some of this with mobility scooters and such, or elderly people who end up in nursing homes for physical therapy and end up never leaving.


For older people doing dishes what often really helps is simply a dishwasher that's installed at desk-height so that nobody has to bend down to use it. A simple (though possibly costing some $$) change that makes no difference to young people but all the difference to elderly folks.


Likewise jobs like computer programming seem pretty 'robot-proof'.

Depends on how you look at it. I see programming as the single most automation-friendly job there is. Even if all you do is add a few lines to your Makefile to run the test suite when you type 'make test', you're still automating a repetitive task you'd otherwise do by hand.

Looking at the bigger picture: programmers are constantly developing new programming languages, libraries, text editor plugins, build tools, deployment tools, continuous integration servers, SCM tools... These things are all designed to save programmers time and if we are successful it means we're increasing our productivity, which should mean we don't need to hire more programmers to do the same job otherwise.

Maybe the last point seems hard to accept in light of the fact that there are more working programmers than ever. Just try to imagine what it would be like to build an app like Uber if you still had to write code on punch cards and batch-feed them into a giant single-tasking computer.


I think there is a similar thing with electricity production, having a fusion generator would give "huge amounts of electricity" until someone thought of a use for all that electricity. There's always something bigger you can find that is limited by current amounts of resources.


Another good one is roads. The old way of thinking was straightforward: "we've got too much traffic, let's build more roads!" Now we realize that building more roads encourages more people to drive. There seems to be practically unlimited demand for roads.


Programming is automation, and that's why it's a job that can't be eliminated by automation.

Every line of code ever written is literally automating something. Yet no programmer ever worries "will this be the line of code that puts me out of work?"


> would you trust a robot to look after your aging parents without hurting them?

Robots should be introduced into caring for the aged as soon as possible.

I can't say this strongly enough: It is a really terrible world for old people who have reached a certain age or level of incapacity.

If you've spent any time with someone caring for an older person at home, or in a nursing home, or in a hospital, there is SO much room for the improvement of people's lives.

home care needs automation.

nursing homes need automation.

hospitals need much much better automation.

Things that people need: help getting someone out of bed. Help getting someone into bed. Help getting someone to go to the bathroom. Help giving someone a bath or shower. Help getting someone some food. Help going outside.

Getting help in these ways is so expensive they basically don't happen.

I could go on and on.


> Robotics still seems a long way off being able to automate many physical jobs - would you trust a robot to look after your aging parents without hurting them?

You're only seeing what's left after centuries of automation/mechanization. Thank of all the tools people doing physical labor everyday use, power drills, various power saws, nail guns, angle grinders, air compressors, etc. These were all done via manual labor not long ago, many within living memory.


Even repetitive, predictable work like making sandwiches to order or picking up litter are tremendously difficult to automate. The tasks seem simple but when you break down all the motions and error conditions there's a tremendous amount of mechanical complexity. I predict that automation will have only a minor impact on those types of jobs by the time most of us reach retirement age.


Sandwiches may become more robot-friendly in their structure, without losing their taste.

Picking up trash is harder. But much of the high-precision picking can be replaced with coarse vacuum-cleaner-style action. The biggest problem likely will be determining what is trash and what is nit.


Machine Learning is augmented human computer programming


Last robot proof job is a fish buying guy? Definitely not. It seems that way because it relies on their vendors to be humans that you need to create personal relations in order for them to give you their best stock. Take those oldies out, make the fish stay fresh in water tanks and slaughter them just before delivery, all done automatically and poof, the job is done by robots. I predict this can be done with current tech nevertheless; automatic fisheries, automatic ships to catch fish, automatic feeding the tanks, automatic delivery.

You wanna know what is the ultimate robot-proof job? Be a politician. That job will never be handled by a robot. Or if that will happen then Terminator style world upon us.


I thought your comment would be a good GPT2 prompt so I tried it. First result:

The fish will have to be delivered right at the time they are needed, since our current technology is geared towards delivering fish at the beginning and then moving on to the next species. At that time, they will have to be fed at the right time, but you don't have any idea how long you might wait or how much time you'll have, and who could possibly get the fish if someone decided to eat the fish anyway.

You will end up with something like this:

In this case, you can even have an artificially intelligent fish, for example, that will be bred so it is much easier to keep the fish alive, but it won't do any better in terms of efficiency of operation. It will simply end up doing less work. A fish with artificial intelligence will just get tired and lose its ability to perform their job.

Another concept is that you could simply use roboticized boats that operate like a large farm or factory, with lots of fish and then moving animals into the water as needed.


Although politicians themselves will probably stay human for the foreseeable future, they have already outsourced their speechwriting and campaign strategizing. The robot ruler won't be an actual robot, it will be a human who uses machines to write their speeches and prioritize their campaign movements.


I agree. In the future will be who is better to write your speech, to the point I can see a politician churning e-mails or what the future will use for comms, from multiple sources, just like offers or like an auction is today.


This seems both highly plausible and deeply chilling. We already saw what happened with electronic voting.


I think it's already chilling that speeches are written by speechwriters who are carefully optimizing for the same things that the machines would optimize for.


So you just listed off several new industries required to replace this guy with a machine.


Yup. Technically all of them can be done by robots. The only reason is they are not is because economics. As long as a job is more expensive by a robot than a human will stay in hands of humans. Once that threshold is passed automation takes over.

But, like I said, while the politics can be done by AI, will never fall in those hands. That's the ultimate robot-proof job. In any country. Regardless of the regime.


I went through yoga teacher training recently. One of my instructors mentioned they think yoga instruction will be difficult to automate because students show up for a personal connection and not just fitness. I think this is an interesting perspective. Especially considering that you can already find instruction on YouTube, yet in-person instruction is still popular.


Hot take: This is true for a lot of jobs.

Very often you actually don't just pay for the job done, you pay for the social surrounding.


Yep, I figured it was true for many jobs. If there is a class of “personal connection” jobs that will be difficult to automate, I wonder if other difficult to automate classes exist.


Also why nursing jobs is more likely to go up in a hospital than down.


People thought that about record stores and book stores too but prices were competitive enough to crush competition. Will be interesting to see how something even more intimate like yoga sessions and if they can be similarly assimilated.


It's also difficult to ask questions to a YouTube video. Even on a forum like Reddit where you might get a good answer, you'll get 10 of them, compared to a person you trust giving you one.


This is the same kind of thing as guitar lessons you Youtube.

There is a lesson for free for zillions of songs on Youtube. The people in the videos are excellent. They might have different approaches that offer different insights.

But none of those videos will have the instructor pause, look through your computer, and tell you your fingers are out of place. None of them will notice which part you are having trouble with and create a new exercise tailored to you on the fly.

Yoga with a good instructor in person is the same thing. You're doing it wrong, you could hurt yourself. You can't even tell you're doing it wrong. The instructor can see it and correct. The video cannot accomplish that.

Some of this stuff it's debatable whether a 2-way video setup can do it. You could have lessons on skype or something but the camera limitations might block this kind of interaction from happening.


Moravec's paradox is still in full effect.

https://en.wikipedia.org/wiki/Moravec%27s_paradox

It only makes sense; brains were doing navigation and control of bodies for billions of years before they were playing chess.


I've seen that page before but I was rereading it and thinking how dumb it is to equate a symbolic integration program with, say, what a mathematician does. This is the exact same mistake being criticized w.r.t. sensorimotor skills - what you don't understand seems easy.


Well, before stuff like mathematica, stuff like symbolic integration is what an awful lot of applied math people spent their time doing. Them integrals ain't gonna integrate themselves!


Before compilers, programmers spent a lot of time writing assembly language, but that doesn't mean a compiler is doing the job of a programmer.


In fact, a compiler is doing part of the job of an old school programmer who only had machine language to work with, just as mathematica is doing part of the job of an old school mathematician or physicist who needed to work out integrals by hand.

It's not "AI" in the sense of actual intelligence of course, more like a labor saving device.

That said, it's entirely possible that letting the computers do these tasks, something important has been lost. It's more obvious in programming when people ship 10 or 300 megabyte "hello world" output when entire operating systems used to fit in 4k binaries. Might also be true in mathematical physics anyway, as they haven't exactly covered themselves in glory for the last 40 years.


Fact check: “It is the second-largest fish market in the world, after Tsukiji market, in Tokyo”

Toyosu Market replaced Tsukiji in October 2018


Although the content of the article is rather interesting, the title is very clickbaity.


Why are they calling this a 'Robot-Proof job' while describing how the guy is training a model?


The models not doing anything, it’s a straightforward e-commerce play. It’s the relationship and curation that matters.

He’s barely using the software.


Cleaning lady is one of the most secure jobs that exist.


Is lady a requirement? Is she providing a pregnancy or being a wet nurse for someone where she cleans?


I live in Africa and around here, people don’t trust men with this kind of work. I have a housekeeper and so do many people, and every single one is a women. You might not be able to prevent them from stealing, but you feel a lot safer than you would with a man.

This has to do with the different evolutionary strategies that women and men pursue. Women tend to pursue a low variance and consistent strategy (which makes sense considering their reproductive cap) while men take a high-risk high variance strategy (which also makes sense given their lack of a reproductive cap).

Women here take stable but probably mediocre jobs: hotel waitress, cleaner, food vender. Men on the other hand are always trying to get a big break and are never satisfied with a job and consequently often jobless. In fact there’s a saying over here translated goes something like this: A man that walks for nothing is different than the man who sits for nothing.


As of 2016, almost 90% of housekeepers were women (Edit - in the U.S.):

https://www.bostonglobe.com/metro/2017/03/06/chart-the-perce...

There are probably many reasons for this, not all of them good. However, one that comes to mind is how it feels vulnerable to let a stranger explore your entire house (likely without you there), and women are perceived as less threatening.


Most automateable jobs are already automated. If there is an increasing robotization of jobs, i m not seeing it. Titles like this are just nonsense.


Lol, this comment will be viewed as “nobody will ever need more than 640k of ram” in the future. The next big wave of automation will be the knowledge workers. Robotics for manipulating objects (manufacturing, transportation, skilled labor) will always have some additional cost to them. But if your job is (data in) -> (data out), it will be much easier, quicker, and cheaper to automate your job.


"Lol, this comment will be viewed as “nobody will ever need more than 640k of ram” in the future. "

No, it's nothing like that. 640k of ram is an absolutely fixed point. "Everything that can be automated" continually increases indefinitely, but the point is that it's not going to discontinuously increase. Or at least, nobody who is concerned ever states a reason that it should.


if by knowledge workers you mean developers, automating their job is their job and they have been doing it since the 60s.

Medical personell -> over-regulated, their pay doesnt fully reflect the work they put in , but the work they don't allow others to put in

Legal professionals -> they control the regulation, they will always pay themselves even if they dont show up to work at all.


> If Warren Buffett orders a red snapper, the company needs to insure that his fish is fresh

A rare grammatical error from the New Yorker?


Maybe he gets his money back if it isn’t?




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: