Your point is spot on but only disproves the likelihood of any coming calamity. Understood in the way you describe, automation making human labor input redundant has been pervasive and relentlessly constant for at least 100 years. This sort of technological progress is precisely what accounts for improved economic productivity.
Interestingly, productivity has plateaued these past few years at the same time we're supposedly entering a golden period for applied machine learning. There are a million ways to explain that away, of course, but it's a reality people should really consider carefully before reflexively dismissing as transient.
In any event, if productivity improvements were to kick into high gear tomorrow we could expect more of the same--increasing displacement but not an overall, calamitous reduction in jobs per se. The increase in the unemployed labor force will open up new opportunities for businesses built around human services, while the reduction in prices for consumer goods will allow people to spend more money on those new services.
Things can totally get worse, especially in terms of wage inequality. But the laws of supply & demand make it almost impossible for a complex labor economy to simply fall off a cliff in the ways that overly imaginative "futurists" and "thinkers" would have us believe. In terms of basic economics, nothing about machine learning is substantially different from historical disruptions.
If you were to teleport somebody from 100 years ago to today, and then to 20 years from now, the relative differences in automation and computer intelligence between today and 20 years in the future would likely be indistinguishable from his perspective when comparing either of those time periods to 100 years ago. And yet without a doubt the contemporary and future economies would be far stronger, wealthier, and even more "equal" compared to the time period he came from. All precisely _because_ of that automation, not despite it.
A lot of that subconciously assumes the doomsday outcome - machines and automation will sweep people aside, and then the reasoning becomes circular.
> At a certain point we're automating new tasks faster than humans can learn them.
That assumes the pace of automation is increasing, but similar concerns have been around for a long time, going back to the industrial revolution. Read Dickens or HG Wells (though a specific cite doesn't come to mind), or look at the 1927 silent film, Metropolis.
But right now businesses can't find enough employees.
> Say it takes 1 year to train a human in something, and 1 year for a robot/AI. well it might take 1 year for the first AI, but copying software is easy. Training the next person takes another year. Even if it's parallelized and you save some time, the cost of training the marginal additional person is was larger.
That's how automation works. Then the people go on to the higher skilled jobs that the machines can't do, including designing, manufacturing, operating, and servicing the machines. Cars made the entire horse industry redundant; calculators and computers put lots of human calculators out of work.
Yet today, with a much larger population, employers can't find enough workers.
If things like that didn't happen, then productivity wouldn't increase and we would be able to afford more shelter, food, healthcare, education, etc.
Up until now, the people displaced from the jobs automated away have always been intelligent enough to offer utility beyond what automation can provide. But there's a tipping point when automation/machine intelligence exceeds humans at which point they offer no utility to employers of any kind. We haven't hit that point for the vast majority of workers, but the early indications are that we very well might.
But looking at the past as a predictor of the future is a very dangerous proposition. We saw that in 2008 when it came to the housing market. Sometimes now really is different from the past. And if we hit that tipping point, it will be much like the posited AI singularity. Vast swaths of humans will become surplus to the most efficient means of completing work. It's important to realize that the people who are talking about this issue as one we need to confront are envisioning a future that is fundamentally different from today or anything that has come before now and they're predicting that for a very specific reason. You can disagree with that reason, but you can't use previous examples of automation-displacing-jobs as a way to dismiss that thinking. Because what's being predicted is fundamentally different from what has already happened.
It appears to me that this time isn't any different. Each major advance in automation has pushed the workforce increasingly into job categories with lower marginal productivity.
When an automation breakthrough occurs a small portion of the population gets better jobs and the rest get absorbed into low productivity/wage jobs. So far the resulting economic growth has meant that these people were still able to afford the neccesities, but sometimes this requires Malthusian forces to kick in and cause massive misery until it is so.
Right now, talk about it revolves around the transformation into service and knowledge economies and the increasing prevalence of bullshit jobs, but each successive wave of automation makes the problem worse. Combine this with the fact that our modern automation advances seem to only be increaing productivity marginally and the overall growth of the economy seems to be slowing down and you've got a recipe for disaster.
>Predictions that automation will make humans redundant have been made before, however, going back to the Industrial Revolution, when textile workers, most famously the Luddites, protested that machines and steam engines would destroy their livelihoods. “Never until now did human invention devise such expedients for dispensing with the labour of the poor,” said a pamphlet at the time. Subsequent outbreaks of concern occurred in the 1920s (“March of the machine makes idle hands”, declared a New York Times headline in 1928), the 1930s (when John Maynard Keynes coined the term “technological unemployment”) and 1940s, when the New York Times referred to the revival of such worries as the renewal of an “old argument”.
>As computers began to appear in offices and robots on factory floors, President John F. Kennedy declared that the major domestic challenge of the 1960s was to “maintain full employment at a time when automation…is replacing men”. In 1964 a group of Nobel prizewinners, known as the Ad Hoc Committee on the Triple Revolution, sent President Lyndon Johnson a memo alerting him to the danger of a revolution triggered by “the combination of the computer and the automated self-regulating machine”. This, they said, was leading to a new era of production “which requires progressively less human labour” and threatened to divide society into a skilled elite and an unskilled underclass. The advent of personal computers in the 1980s provoked further hand-wringing over potential job losses.
and based on a simplistic understanding of how automation impacts the demand for labour.
Automation reduces costs, and this increases consumer spending on more difficult to automate goods/services. [2]
That has been the pattern for 200 years, and why wages today are 20X what they were in 1800.
Unfortunately, ideas like yours are very common [3] and lead to terrible government policy.
I'll tell you why this time is different. Each time we had growth in technology that had the potential to displace a huge number of workers that technology also enabled a massive decrease in the costs of transactions (railroads, motor vehicles, the internet, etc). This allowed a massive growth in the economy which enabled these displaced workers to find new jobs. Also consider that the growth in technology didn't eliminate jobs, but it shifted them. The technology of old still required a mostly proportional human energy input to produce output. New technology decreased that proportion, but it didn't eliminate it.
This last point is the difference that this new wave of automation has compared to the waves of past. It is unprecedented in human history to be facing a future where human effort can be multiplied by such enormous multiples that the value of labor approaches zero. This coming automation revolution would have to simultaneously grow the economy to greater proportions that automation reduces the cost of labor. I don't believe this will ever be possible, but definitely not in the time period that a properly developed AI and robotic technology could corner the market in labor (once developed, less than a decade). Society is not prepared for what is coming, and it will be disastrous.
Exactly, we’ve been automating the living crap out of almost everything for hundreds of years, but employment has never been higher. Automation leads to higher productivity, cheaper goods and services, increased need for technical skills, thus higher wages and higher demand.
Of course there can be some painful periods of adjustment, but that’s often caused by misguided policy trying to hold back the tide and delay the inevitable.
Ultimately yes, full no holds barred human level AI may well render human labour obsolete, but we’re a very long way away from that.
I know some people think LLMs are close to that already, but no, not even remotely close. I do believe strong AI is possible and maybe even inevitable. They’re a huge step forward, and are easily the biggest advance towards strong AI in my lifetime, but these things are just tools.
I'm not at all assuming all technological progress is an unadulterated good - in fact, nothing is, everything would have a downside if it has an upside. All I'm saying is it's still bound to happen and we can do nothing about it.
It's not the first time that 'entire classes of skilled jobs' have been eliminated. It's the natural side effect of automation and has happened with every generation of new products. Industry automation might have taken millions of jobs away so far to give us better products faster. Tomorrow my job as a usual developer might be gone to these AI tools (though it's a far fetched thought), and I have to be prepared for it, and re-skill/up-skill myself if I ever sense that day is about to arrive.
This is exactly wrong. Automation happened already, 200 years ago. It upended society. It was called the Industrial Revolution. Things changed dramatically, and nothing remotely of that scale is occurring now.
We are suffering today not because of technological progress, but rather the contrary: the rapid technological progress of the 20th century that brought tremendous economic prosperity to humanity has finally come to a grinding halt. Let's stop denying this. The stream of lifechanging breakthrough inventions of the 20th century, from A (antibiotics) to Z (zippers), have ended. As a result, we now suffer from secular stagnation. [1]
This is why people can't get jobs. The answer is Keynesian spending, and the most fair kind of Keynesian spending is basic income. Call it a "Keynesian dividend".
1. Speed matters. Going from 100% to 2% over a century is one thing, doing it in 5 years is another. Which isn't to say that it'll necessary be that fast, just that if it is (and there are some reasons to think it will be) that it will have serious consequences for human welfare.
2. People like to point at the industrial and digital revolutions as examples to say "hey look, you can enhance productivity without killing jobs - the jobs just move around". And that's fine, they're excellent examples of that phenomenon. However, two prior examples is not exactly overwhelming evidence, and there are at least a couple real reasons to think that 'this time is different'. Specifically, that AI is getting really good at mimicking humans for particular tasks. We certainly don't have general AI yet, but what we do have can get really good at a whole lot of low-skill, low-education jobs, and a few high-skill high-education ones too.
When the industrial revolution happened human jobs persisted because humans could use judgment and learn new things. It was too expensive to automate everything that could in principle be automated, because your processes might change tomorrow. New manufacturing tech is much better at reconfigurability, so that defense is likely out. New AI techniques are able to replicate human judgment in many scenarios now too. Not all of course, but many. If you eliminate all jobs that involve a human doing a particular physical task, you're not left with a whole lot for third world or low-education first-worlders to do.
Now, all that being said, would I be extremely surprised if automation came along and labor markets simply retooled around it and everything was fine? No, not really. But I wouldn't be surprised the other way either, and I think it's something we should take pretty seriously.
In 1850, 60%+ of the American labour force were farmers and over the next 100 years their jobs were automated out of existence by range of more efficient 'robots'. This was no-doubt harmful for many specific farm labourers, but didn't 'screw over the economy'.
I hear that argument all the time, but I wonder if it takes into account the ever increasing pace of technological advancement.
Your entire viewpoint is based on the idea that there will always be jobs no matter how advanced ai and automation becomes. This is the idea that needs to be addressed, not the shrubbery surrounding it. It’s meaningless to say that new jobs have always been found and that rapid advancement in technology has not seemed to doom us so far. That’s like saying we haven’t run out of oil yet so why would we ever run out. The number of tasks that a human can do is finite. When machines saturate that set of tasks, something really bad will happen. But even long before, there will be massive problems as we approach it.
I already know you won’t be convinced by what I’m saying. Just respond with your strongest counter-argument to my central point about the finite nature of human jobs, or in other words what will happen when machines can do everything we can. Just give me your strongest counter-argument about that. I will then explain why that’s wrong, and we can continue until you see that you’re wrong. I will be extremely patient about it.
I think you're right, that does seem like a possibility. It seems unlikely to me that our wage-paying jobs will be phased out by automation that rapidly, especially if you consider the whole global economy. Of course, I could be completely wrong -- I guess a true Singularity could invalidate almost all labor in a matter of years or even months, depending on what form the AI takes and what it invents.
From a historical perspective your reservation is totally fair. But I think the issue is a lot more complex than just looking at the impact automation has had in the past. Here are a few points worth mulling over:
1. Automation in the past has been good at increasing worker productivity, and making new avenues of work possible. This is true going forward too. But each new generation of technology isn't "something we've seen before", it's a new thing with new consequences. Technological development isn't cyclical, so estimating the impact a new technology will have on the impact previous technologies have had is a poor model. This isn't to say the consequences will be bad, just that they're hard to anticipate. For the most part, the historical perspective is probably right though.
2. Automation in the past has been very effective in improving worker productivity, but developments in AI and robotics are looking at ways to supplant workers (i.e. electric cars don't improve our ability to drive, it removes our need for drivers). While in the broad sense this trend is good, and people will over time shift into new industries, it is going to be disruptive. Timelines will have a big impact on the shock. New industries won't spring up over night.
3. The timeline for this level of automation is much shorter than previous automation trends. The shift in agriculture happened over generations. The shift caused by driverless cars will likely happen in less than a decade. Add to that the efficiency of market pressures we have today - once one business is able to shift entirely to an autonomous fleet and save money over their competitors, all their competitors will have to follow suite to remain competitive. Entire industries could be displaced, and those workers will need to move somewhere.
4. This is more hypothetical and longer term, but it gets at what I think is the general fear around automation. Imagine we develop the ability to automate any unskilled job (whether through broad automation improvements or development of an actual general-purpose automaton). The primary factor in whether a business would choose to employ that automaton over a human is cost. The automaton is a once-off fixed cost, whereas a human is an ongoing cost. Once the cost of the automation is lower than paying a salary, humans will no longer be employed in that role.
5. Following from this, if it's the unskilled jobs that get automated, where do those workers go? At this stage, even if automation is creating new lines of work, why wouldn't that work also be automated? Basically, once we automate unskilled work, we never need unskilled workers again. In order to find work they'll need to skill up, which takes time and money. And if it takes a year to train a worker to a level that they're a net benefit, why not invest that money instead into automating the skilled work too?
We are now going through a major economic disruption, as human labor is being replaced through technology and automation. Yes, in some areas, as described here, the replacement is more like assistance than elimination. But, in time, more and more and more of work will be done by machines. Machines will even manufacture and repair machines. At that time man will enter a new era - a post-capitalism era where we will be free to do what we want to do, not what we have to do. It is like being retired with enough money. Between now and then, things will get worse. Politicians and the rich are misdirecting our attention. We are told it is because of offshoring - whereas only 14% of all job losses are due to that, whereas 86% of all job losses are due to technology/automation replacing human activity (stats are provided by expert with http://yourhomeworkhelp.org/do-my-statistics-homework/ ). The investment class is reaping the benefits - while people compete for fewer and fewer good jobs. The costs of Education, Healthcare and Housing continue to outpace wage growth for most. Many troubles of today's society will get worse until we get a handle on managing the transition. Still, as stated here https://www.nytimes.com/2017/03/19/technology/lawyers-artifi... and as I believe AI cannot replace the actual human being and I hope it never will
I gave that last item a bit more thought, and I think that may be the mitigating factor in all this. You can only automate so much / put so many people out of work, until it becomes unprofitable to automate anything else because the market will have fallen so far.
So rather than a single huge mass event where automation destroys our fabric of life, we'll have swings back and forth from overautomation to overemployment. Each swing will give us a chance to re-evaluate and rebalance our priorities.
Essentially that's already happening and has already happened over the last ... couple centuries really. I guess there's a chance that the pendulum could suddenly swing so far in one direction that it's irrecoverable, but at least I don't think one can say it's inevitable.
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.
I should have said today's big attempts at further automation.
> We have plenty of kiosks, self-checkout, automated fulfillment and logistics, etc happening. We also have plenty of higher skilled jobs having portions automated. Productivity is way up in the last 50 years, and that's through automation and mechanization.
We have indeed accumulated tons of this stuff over that time, and that's good! But the more labor id frees up, the less incentive it is to continue down that route.
Like, why buy another kiosk, if there is a gazillion people clamoring to be your cashier?
The only way to keep constant the financial incentive for further automation is to keep labor markets tight, which we either do by bumping up demand, or reducing labor supply.
My point is a lot of people talking about a "jobs apocalypse" seem to assume the technology will just happen no matter what, but that's not true. Stuff gets stuck in the research phase if it's not profitable to develop. The supply of "potential automation" doesn't create it's own demand for actually deployed automation and this is even worse, because the resulting stagnation obscures the root problem.
Indeed, productivity growth (the rate of automation) is fairly low now. A lot of the big stories about automation killing jobs started popping up after the financial crisis, ignoring the actual reasons for the job loss (the financial crisis that had just devastated the global economy) in order to push a particular viewpoint. Though it was claimed that automation had permanently destroyed certain jobs and was about to destroy many more (and that the rate that this was going to happen was about dramatically), labor force participation has actually been going up since then.
This is starting to remind me of peak oil, where it's become an article of faith among some people that there's a disaster right around the corner, and no amount of failed predictions or facts pointing to the contrary are going to convince them otherwise.
Interestingly, productivity has plateaued these past few years at the same time we're supposedly entering a golden period for applied machine learning. There are a million ways to explain that away, of course, but it's a reality people should really consider carefully before reflexively dismissing as transient.
In any event, if productivity improvements were to kick into high gear tomorrow we could expect more of the same--increasing displacement but not an overall, calamitous reduction in jobs per se. The increase in the unemployed labor force will open up new opportunities for businesses built around human services, while the reduction in prices for consumer goods will allow people to spend more money on those new services.
Things can totally get worse, especially in terms of wage inequality. But the laws of supply & demand make it almost impossible for a complex labor economy to simply fall off a cliff in the ways that overly imaginative "futurists" and "thinkers" would have us believe. In terms of basic economics, nothing about machine learning is substantially different from historical disruptions.
If you were to teleport somebody from 100 years ago to today, and then to 20 years from now, the relative differences in automation and computer intelligence between today and 20 years in the future would likely be indistinguishable from his perspective when comparing either of those time periods to 100 years ago. And yet without a doubt the contemporary and future economies would be far stronger, wealthier, and even more "equal" compared to the time period he came from. All precisely _because_ of that automation, not despite it.
reply