Tech companies should stop pretending AI won’t destroy jobs
technologyreview.com
technologyreview.com
Saying "find the jobs that AI can't do" like it's this obvious thing; we haven't the slightest clue what AI can't do in five years. Practically yesterday Go was the go-to example of what computers couldn't do, and now it's the go-to example of how they kick our ass. Well-informed predictions for widespread self-driving car deployment range from "in five years" to "half a century at best." And the jobs that aren't threatened by AI, and that are accessible to most people, are threatened by much "dumber" kinds of automation. Human-facing "empathetic" jobs are often considered safe, but what positions actually need a human, anyway? People are happy to do a bit of extra work to get "expert advice" from a database when it comes to travel, law, or finance; stores are happy to let customers use automated checkout rather than talk to a cashier. Dextrous, highly-physical work is still safe (pending advances in robotics, which seems to move a bit slower), but only until someone figures out a way to offload the dexterity-requiring bits onto a factory, or render the whole thing obsolete. And finally there's second-order effects of the resulting economic turmoil; if most of the middle class can't count on keeping their jobs, not enough people are going to be ordering fancy lattes to pay for your role as a barista. Planning a relatively low-skilled career (anything less than a four-year technical college degree) is walking a minefield.
Us highly-educated technical generalists are probably fairly safe, but not everyone wants to study for 6+ years after highschool- and regardless, they could expect to have to relearn all the applied skills in their chosen field when the tool they originally studied gets replaced by something completely different and orders of magnitude better. The old model of "learn a skill, get a job applying that skill, eventually retire" will no longer reliably work on any point of the scale.
https://foldimate.com/ https://www.theverge.com/2018/1/10/16865506/laundroid-laundr...
(and yes these are buggy and expensive now, but 5 years?)
I totally agree with you. But even in the present that's a tough sell, many people derive all or most of their self worth from their skill and their capability to generate income. It's the second question people will ask a new person they come in contact with, right after 'what's your name', and long before 'what are your hobbies' and 'do you have kids' or 'are you married'.
And it's already begun, no need to add your 'before that happens' disclaimer, not all jobs lost to automation have resulted in successful transfer to a different job.
The job market will get squeezed from all sides, on the high end because of the pension requirements it will get expanded to keep experienced and older people away from their pension, on the low end you will see an increase in the number of qualifications required for a particular job and automation slashes like a giant comb across the whole thing to remove those jobs that it can now absorb on a regular basis.
We need to find the jobs that AI can’t do and train people to do them
The author raises and then dismisses UBI in a handful of sentences. Then he asserts that we need to be inventing AI-proof jobs -- not because they're useful and worth doing but just to avoid having to change.
The changes we have coped with as a species are incredible. From hunter gatherers to agrarians, to the development of civilizations with laws, to the industrial revolution, etc.
I think it's an incredibly pessimistic and narrow view of humanity to presuppose we need to be making busy work for people.
Of course corporations and politicians should be approaching this problem honestly. But we should also approach it open-mindedly.
So no, people don't need busy work - and that's the challenge.
Electronic computers 'destroyed' jobs when they replaced human computers.
Motorized street sweepers 'destroyed' jobs when they replaced people with brooms sweeping debris off of streets (except in Beijing where, as of 2015, this is still a thing).
Firearms 'destroyed' jobs when bowers no longer had any reason to mass produce bows and arrows.
People will find a way to stay busy. The past is literally filled with examples of a new technology replacing people, and those people eventually found other things to do. It's not hard to see, in hindsight, how many of my silly (but real) examples above translated into entirely new industries.
Pardon the harshness here, I feel like the only people this is truly a surprise to is either folks that ignore the past, or are incapable of understanding the past.
You are implying that this will happen automagically, and there is no cause for concern. I disagree. People will find other things to do, but if we want there to be jobs in 20 years for people who have lost their jobs to AI, we need to get to work now.
To use an example to illustrate my point, people have always found new sources of energy. We have never ran out of energy, nor faced extreme shortages as old resources dried up. Does that mean we should ignore any calls to action for investing in alternative and renewable energy sources? After all, why should we worry about and waste resources to prevent a thing that has never happened? Humans have always found a way, I'm sure in time, when oil and gas and coal dries up, we will find something else. Right?
Things work out in the end because we make them work out. Waiting until the disaster hits and then starting to think about possible solutions is not a reasonable approach. If we want to have alternative energy resources by the time we run out of fossil fuels, we need to start working towards it now. Same if we want to have alternative jobs by the time the current ones become obsolete by AI.
Nor is it a sensible approach.
Excellent points. Whenever thing have been solved, it is because humans did something about it. Not because we sat around waiting for random chance or historical precedents to solve things.
Saying not (of) "it will be different this time", i.e. "it will be the same this time", i.e. your claim, can also turn out to be false.
>bowers
https://en.wikipedia.org/wiki/Bowyer
>Pardon the harshness here, I feel like the only people this is truly a surprise to is either folks that ignore the past, or are incapable of understanding the past.
No need to ask for pardon. But what happened in the past is no guarantee of what will happen in the future. See Black Swan etc.
It's like being a horse at the turn of the last century saying well we've always had horse jobs throughout various advancements in horse-labor tech, this internal combustion engine should be no big deal we'll find horse jobs.
We all know how that turned out for horses, now we're pushing closer and closer to the precipice of being the horses at the turn of the last century. Sure it'll be impossible to tell whats going to be the straw that breaks the camels back but we should be ready for each straw to break the camels back, and not assume it won't.
We have the opposite situation with humans. Individual humans retain their reproductive rights. Most of them don't decide to reduce the supply to meet a reduced demand. For a time, China did impose a restriction on reproduction, but that just resulted in many (mostly female) babies being killed or abandoned.
We will find something to do with humans because we'll be forced to. Humans are exclusively more qualified for existence.
All of your examples caused the same misery, then lots of people died,some adapted like you said . Nobody is arguing to become Luddites, but I do see the need to acknowledge we need to handle the transition ( if you believe in solidarity, Christian or other religious values or in general care about humans ) or be honest when we do not care about peoples misery and death.
If you just take the first step - self driving cars - and remove all the jobs that currently have a driving license as a marketable skill away from people that only have that particular marketable skill (and that's quite a few of them) then 100's of millions of people the world over will instantly be unemployable.
And that's just one step. That does not translate into 'they'll go and do something else', quite a few of them will end up in poverty, commit suicide or worse because they've been ingrained with a mentality that tells them that if they are not able to contribute that they are a drain on society, and that's for those societies that actually have a safety net. For societies that do not a life of crime or begging is pretty much their only option.
For all the smarts that the HN crowd possesses quite a few of us are categorically incapable to see the world through the eyes of the people whose lives are affected in a very harsh and direct way by our creations.
Just stating that 'people will find a way to stay busy' is easy for someone who never has enough time in a day to read all the books they want to read, build all the stuff they still want to build and who probably scores in the top 5% income wise. Which makes it very well possible that you'll end up in the bucket of the haves rather than the have nots.
The rest of the world will likely not be that lucky. And as one of the haves it should worry you. See also: the French Revolution.
MAYBE driverless cars become a reality and quickly at that. MAYBE that makes drivers in the west redundant just as fast. But worldwide? I think people underestimate how inexpensive labour is in places like Bali, let alone Kenya.
No one knows what will happen, or what new industries will be formed as a result of these changes, and the warnings are all good and well, but there is much doubt in all directions, and crazy policy implemented to avert a disaster may be just as bad as doing nothing.
That's why this whole new level of automation is so scary: no body has any clear idea what will happen, and the policy shift is really, really unclear.
> No one knows what will happen, or what new industries will be formed as a result of these changes
No, but we can make some pretty good educated guesses:
Anything that does not require a skill that can't be automated will eventually go away. How long the transition will take will be a major factor in whether or not it will be a smooth one or a terrible one. I hope it will be a century or more, I fear it will be a decade or less once the keys to AGI are unlocked. Worst case it will be shorter than that. In the last two cases: buckle up.
> and the warnings are all good and well, but there is much doubt in all directions, and crazy policy implemented to avert a disaster may be just as bad as doing nothing.
That sounds like a global warming piece.
> That's why this whole new level of automation is so scary: no body has any clear idea what will happen, and the policy shift is really, really unclear.
Yes it is. As far as shocks are concerned this will be one of the biggest that humanity has had to endure so far.
1. You are assuming that what happened in the past will continue to happen in the future. Maybe it will, or maybe this time things are different. There are reasonable arguments that make the case that this time is different. The difference is that computers are encroaching on the one point of difference that man has always had on machines - his brain.
2. You are ignoring transition costs and times. Even with past mechanizations, there was immense pain and suffering and disruption that went on for decades. For neoclassical economists it will be OK in the long run, in equilibrium, when there will again be full employment and fairies will dance gaily on colorful toadstools.
Back in the real world people suffer and die.
For a comment so seemingly proud of its historical knowledge, this seems a very bizarre claim.
People as a whole could be said to have adapted, but the affected individuals very often didn't. The suffering was immense. Modern social safety net and the welfare state are directly related to the abject poverty resulting from the dramatic and often cataclysmic urbanisation of the industrial revolution.
So yes, humanity will survive and adapt. It won't be the end of the world. But avoiding analogous problems to the concomitant evils of the industrial revolution is exactly why we need this discussion. Shrugging one's shoulders at the problems of history, because a new equilibrium arose, seems to me to betray the real lack of historical understanding.
Why do you say this? Plenty of unpleasant experiences have occurred via industrialization in the past. Industrialization can be good for society overall and result in new jobs over time, but for a given set of workers today, they're looking at losing the jobs they have no and seeing no replacement for it, as it is unlikely that any of the newly created jobs would be willing to take them in.
It's an illusion to think that everybody will be happy being pensioned off at 23, especially if that means that they will be without income.
This shift will move a lot of wealth to the top 5% or so and will leave the remainder in serious trouble.
"everything that can be invented has been invented." - Charles H Duell 1899.
The naivety is in assuming that there are simple limits to what AGI will and will not be able to do.
It's not an evolutionary change, it is a revolutionary change where the world as you see it today will have very little resemblance to the world the way it will be afterwards. If that change happens too fast there will be a lot of blood spilled.
Ignoring all this is one surefire way to bring about one of the worst variations on the theme - assuming it is inevitable, which the jury is still out on but the first steps on the ladder are being taken as we speak.
There was a good article on here a while ago about how safe your job is:
https://news.ycombinator.com/item?id=14443606
Now zoom out to the rest of the world, take a decade or two as the timeframe and multiply by the likelihood.
The first consequence of major technological revolution, the post industrial era enabled the biggest wars human history had yet seen because our societies where technologically advanced but spiritually depraved. Modernity gave us amazing technology but also totalitarianism we had never seen before.
Yes, doomsayers are always numerous but just because people are crying wolf a lot doesn't mean that wolves aren't real. The image of a society in which hundreds of millions of people are devoid of purpose, live on subsistence social welfare and are at best employed to keep them occupied is a very scary one. And it might very well happen.
It’s worth reviewing the actual list for yourself to see what these common jobs really are. And thinking about at what point you would rather be dealing with a robot (let’s call them that) than a person.
Truck driving is #15. It gets a lot of attention because it’s a good paying job for a high school education.
Taxi driver is #173. Seems low for all the attention that Uber’s disruption of that industry gets.
1. https://www.bls.gov/opub/ted/2016/mobile/retail-salespersons...
A post-AI economy could easily be an economy of artisans, one where it’s easier to afford the more basic necessities of life but where human work and effort still has value.
"According to a Rockefeller Panel report commissioned to take an in-depth look at the financial realities of the performer's life:
"The miserable income of the majority of performing artists reflects both a shortage of jobs and the brief duration of employment that is available. In all except the small handful of our major and metropolitan orchestras, musicians earn an average of only a few hundred dollars a year from their professional labors. During an average week in the winter season, only about one-fifth of the active members of Actors' Equity Association, the theatrical performers union, are employed in the profession. Of the actors who do find jobs, well over half are employed for only ten weeks -- less than one-fifth of the year. For most opera compnies the season lasts only a few weeks. The livelihood of the dancer is perhaps the most meager of all."[1]
Now, that's today, with relatively few artists. Imagine a future where there's a flood of artists from all the people who lost their jobs to AI (assuming they all have some desire, ability, and dedication to actually try to make a living as an artist.. very unlikely, but such is supposedly the main hope for a future for these people). Now the wages of artists will be even further depressed as there will be a glut of supply (much like the glut of "photographers" now that everyone has a camera in their cell phone).
Then it has to be asked: who will be buying all of these artisonal products and services from the newly unemployed masses? There's only so much art the relatively well-off are going to want to buy, only so many plays they'll pay to see, and so on. Demand would have to rise tremendously to absorb all the additional supply if the already meager wages that artists earn aren't going to fall to virtually zero. Where is all that extra demand going to come from? The US is already not a place where art is particularly valued. There's no indication that that's going to improve with the advent of job-replacing AI.
[1] - From "Staying Sane in the Arts" by Eric Maisel
A poorer home in the future might have a cheap AI-made TV, cheap AI-made cookware, a fridge full of groceries delivered either by drone or self-driving car, cheap mass-produced IKEA furniture assembled by a robot, all of these things “cheap” by cost but still higher quality than the mass-produced goods of today, but maybe the one luxury might be a handmade rocking chair from a local woodworker, who lives in similarly comfortable standards and spends his days building and selling chairs and canoes and cabinetry and whatnot, none of which is necessarily materially better than the stuff manufactured in drone factories and shipped across the world in drone ships and drone trucks, but people will pay extra for the small luxury of having their own special handmade things sometimes.
Also, because the necessities of life are so cheap, the woodworker can invest a big chunk of his woodworking income in index funds and have a share of the AI-generated wealth.
People might be paying a premium for hand-made trinkets now, but if there's a glut of supply from all the new artisans, those prices will plummet unless a huge amount of new demand for these artisonal goods materializes.
Sure, these newly unemployed people might themselves occasionally be able to splurge on some hand made trinket, but it's very unlikely that they themselves will be providing enough income to sustain each other. As now, the vast majority of purchasers of artisonal goods are likely to be the relatively well off, and there's just not that much demand among them to absorb the new glut of supply, so artisonal wages will necessarily plummet.
That's all assuming that the newly unemployed will will want to and somehow magically be trained to make furniture, jewelry, hand-made ice cream or what have you. Doing that to a high enough quality that people will actually buy and value what you make to outcompete all the other artisans making similar stuff is not going to be easy.
Also, it's far from clear that the cheapness of manufacture of AI-made goods will result in lower prices for the consumer. Prices consumers pay might remain the same and the extra profit be pocketed by the manufacturers.
To your final point about investing in index funds, it's far from a certainty that the index will continue to rise in the future. It could fall for any number of reasons, in which case even those fortunate enough to be able to afford to invest in the market will suffer.
Policy will be a huge wildcard and also criteria for long term success.
I think we over estimate where and by how much automation will strike. Poor to middle income people in rich countries are most at risk. Question is, can we pivot the country to find gainful employment for the dispersed peoples?
Nothing really impressive has actually come out of AI yet except running old AI on bigger databases and faster hardware.
What you will find is that radiologist and other highly specialized jobs actually are much easier to replace than say a cleaning lady because it's a job that requires a lot of different skills.
AI doesn' have to be impressive or even come close to being general AI to be a danger to humans and the jobs that require the most of a human are often not the best paid jobs. The will be even less true if humans have to flock to al be cleaning ladies/men.
Minimum wage is probably the worst form of income redistribution we use.
not allowing people to have salaries competitively priced vs machines will cause most of the swift pain AI will bring.
please, replace minimum wage with more welfare or better yet negative income tax.
Of course, some jobs will die regardless. But minimum wage greatly exacerbates this.
Previous comments have suggested this trend will not last. Be it that AI will reach a point where it is capable of taking on any new job that we throw at it, or that we will simply run out of new ideas for new kinds of jobs at some point in the future. That is the context being discussed above. In that case, your options are: Compete against automation on price, or go without work entirely. Minimum wage prevents the former, leaving only the latter as an option. The parent is saying it is less ideal to have a $0 income than it is to have a less than minimum, but above zero, income.
But that all hinges on the assumption that work will dry up. The opposite has held true for centuries. It seems highly unlikely that this time is any different. I sincerely doubt we have reached the pinnacle of human achievement already.
When streets have 300 beggars on it, good luck making any money. I live a few blocks from skid row.
I haven't heard any reports of 14% of adults dying off each year due to not making enough money, and that would be a pretty big news story, so it seems that they have somehow managed to survive. And while the data indicates that some have chosen $0 (~4%), and maybe a life of begging and crime (the data is not sufficient to see that correlation), it seems some dollar value greater than zero is still the preferred option for most even when their earning potential is limited.
Humans may want to sharpen their pencils and outcompete automation on price, but are prevented from doing so. Robots never work for free. Capital, operational, and maintenance costs can be quite expensive. However, the cost of automation amortized on an hourly basis may be below the price of hiring someone at minimum wage. As such, humans are legally forced to find something else, or go without work entirely.
Given that automation has historically created far more jobs than it replaced, you can argue that it is good thing that people are forced into more useful roles by not allowing them to compete with machines. But if you believe that automation simply takes away jobs for good, as many people do, then minimum wage will actually result in even lower wages for these people than they would have without minimum wage.
The automated dishwasher in my kitchen relieves human dishwashing services to the tune of an hour or two per week. Minimum wage where I live is $14/hr. That is, on the low end, a little over $700 per year. Would I be on the hook for the $700 each year, minus the current operating costs of the dishwasher, in order to have it operational in my home?
And should the taxed value be a straight up calculation like that? Humans are quirky and unreliable, and I may not be comfortable with having a stranger in my house to clean up after me. As such, I may be willing to pay quite a bit more to have the reliable and comfortable automation over hiring a human. Would the tax need to account for that disparity to allow humans to still have a competitive edge?
I'm not convinced this hypothetical situation presented is realistic to start with, but that is well beyond the topic at hand. The constraints are what they are.
Minimum wage is, by all accounts, a success. It prevents employers from creating jobs that prevent a person from ever being successful. And we know that it's good and successful because we have the data to show that it does indeed raise the quality of living to a "barely first world" standard for many people. Seeing this on HN:
> not allowing people to have salaries competitively priced vs machines will cause most of the swift pain AI will bring.
is startling to me. Who does the OP think this person is that can live on $5/hr or lower? Minimum wage _as it is_ is arguably too low. It's sad that this illusion of "if you lower minimum wage, people will work for the lower wage" seems so pervasive, because it's thrown out as if minimum wage employees are trying to be competitive. If you've eaten twice in the last two days and you're about to lose your car, you'll work for almost anything if it means keeping your head above water for a couple more days. Minimum wage is a solution to a human problem, not a business problem. If a job can't be performed competitively for a survivable salary, employers shouldn't be allowed to say, "well if you don't mind not having heat in your home..." and should instead be forced to use automation. Yeah, it sucks being laid off, but there will always be more minimum wage jobs. If you take away minimum wage, now you're left with even fewer reasonable prospects of a job that pays a survivable wage.
There are better alternatives to minimum wage that don't have the same problems with AI.
one example:
https://en.m.wikipedia.org/wiki/Negative_income_tax
People think wanting to get rid of minimum wage is to be against helping poor people; it's actually the exact opposite. Being against minimum wage is to be against the problems it brings - not being against the poor.
But, given the context of discussion, the only other alternative is to have no wage at all. Being able to buy some things allows you to live better than being able to buy no things. We're not talking about the real world where automation creates more jobs than it destroys, we're talking about a hypothetical world where your only choice is to go directly head-to-head with automation. You have no hope of winning if the law forces you to charge more than the competition.
> It prevents employers from creating jobs that prevent a person from ever being successful.
You're slightly off on that. Minimum wage prevents employees from competing against each other once the price reaches the defined minimum. It's to protect you from other workers, not the employer. Minimum wage exists precisely because people have proven over and over again that they will sharpen their pencils and figure out how to do job for less than you in order to get the job over you.
If there was nobody to undercut you, employers would be forced to pay you whatever you want. This is why software developers, who are limited in availability relative to the jobs available to them, are generally able to charge significantly larger amounts of money. Jobs at the minimum have such an overabundance of labour that they're all fighting for the limited number of spots, and thus there is always someone willing to work for less to get the job.
> Who does the OP think this person is that can live on $5/hr or lower? Minimum wage _as it is_ is arguably too low.
Although it seems kind of silly to compare the needs of our current situation to some hypothetical future where everything is automated. If humans were doing all the work that automation currently does today, the current minimum wage wouldn't just be arguably too low, it would be but a tiny fraction of what would be necessary to survive. Food, for instance, would easily be 10x more expensive than it is now if we didn't have all of the efficient automation keeping the costs down. Imagine trying to live on the current minimum wage while paying an order of magnitude more for your groceries. With automation taking over in this hypothetical future, the cost of living will have no relationship to what we're accustom to in this current reality.