How's this any different than say, tractors, or the mechanical loom? After all, agricultural employment went from 90% in the past to 1% today.
Not at all. I doubt LLMs will result in a 90x drop in the overall labor workforce. The agricultural shift was likely greater than the shift due to LLMs will be.
That’s fine, but one of those assumptions has to be the case for your statement to be true. If they meet or exceed all productive human capacities at lower cost, are not stopped by regulations or some kind of near insurmountable “exponential cost of intelligence”, then this is completely utterly different than the agriculture shift.
My generals observation about people like you is that you assume “the future form of ai is just a chat bot, like today” and that is just not the case, and it’s not what anyone is worried about. Many of us are “playing” with real agents, grafting together agentic memory systems, kicking around early experimental harnesses, seeing what kind of self learning loops we can hack, perfecting evals, wiring in eyes and ears and a heart-beat to these things. And those of us who are often take a look at the Frankenstein result and go “ya, this could completely replace us with enough iterations”.
We should at least be scared enough to seriously consider the possibility that in the future there is no productive use for human labor, only capital.
Oh, as a SW engineer, I assure you I am scared. My profession will be one of the most impacted ones.
My point is that a lot more than 10% of the labor that is done out there involves things that require physical work, and that is a tougher problem to solve than pure reasoning. I'm not saying the changes won't be drastic - just not a 90x drop.
I agree in some sense, I think the runway on human physical labor is measured in decades. But it’s likely not infinite, and likely not 100 years. I expect my children will at least see “the end” in their lifetimes.
Human physical labor also has a ticking clock, albeit longer than software engineering.
Of course again, this is if we don’t put up or find unknown-unknown walls on machine intelligence. I think there is a pretty high chance we just make sufficient machine intelligence illegal for a very long time, at least until labs can fine-tune models enough to be smart enough to replace humans but dumb/lobotomized enough to not make a bio-weapon.
Regarding US agricultural labor displacement.
* It happened over a period of 200 years or so in the USA. [0] That's a key difference.
* Starting in the late 1800s manufacturing rose to a peak of 38% in 1944. [1] This absorbed a lot of the available labor, often at better rates of pay than farm work. It's a common pattern in industrializing nations where manufacturing absorbs labor freed up by more productive agriculture. Manufacturing labor is no longer growing, so that cannot help with employment.
That's not to say it was pleasant for all concerned. I would argue, however, that black swan events like the Dust Bowl caused more disruption and trauma than the steady displacement of farm labor by technology.
[0] https://u.osu.edu/beef/2022/07/06/the-history-of-american-ag...
[1] https://humanprogress.org/trends/the-changing-nature-of-work...
Edit: clarity
The pitch for AI is that it's affordable at the insane valuations because it replaces labour.
It takes work out of the labour market entirely — fewer salaries means more money can be freed up that can go to the giant intelligence tap.
Not just some sectors — really all non-manual work sectors at once. Isn't that what the e/acc guys were open about at the beginning? Learn AI or you won't have a job?
Sam Altman was so open about this that he funded a UBI study.
source?
Farmers went from working outside at a stable work pace (and in many cases farming a small patch of their own land as part payment, so eating at least functionally well) to being forced out of their farming work by the second agricultural revolution (leading to the Swing Riots, Tolpuddle Martyrs etc.) and to living in cramped industrial slums, working in appallingly dangerous and polluted factories, long hours, terrible food, toxic chemicals, severe health issues.
Subsequent infant mortality in industrial area families was about twice the rate in industrial areas as it was in rural areas because of appalling living conditions and poor food.
It's the underpinning story of the second agricultural revolution and the industrial revolution.
An interesting link here:
https://www.nationalarchives.gov.uk/education/resources/vict...
But this is well-studied history. The industrial revolution did not liberate the poor until labour law changed to stop them being expendable; living standards took the best part of a hundred years, until as late as the early 1900s, to return to a level where people were as healthy as they were or to live as long.
But even if this were not true it’s still not a working analogy for AI, which is going to eliminate employment, not just job roles. It’s the whole pitch for AGI.
1. Where's the hours data you're citing?
2. My whole point in the previous comment is that there's more to being a medieval peasant than just plowing the fields. If during the industrial revolution you spent more time in wage employment, but then spent less time on household tasks, that would be captured in the hours data as simply more hours worked, because the latter isn't accounted for in the hours data at all.
Second, work being split by gender does not matter here. Women are, by definition, people too. And weaving, sewing, candles crafting were all literal necessity. A weaving woman would sell or exchange results of her work if she had an excess of it. They were not bored SAHM hobbies they way they would be now. This was economic activity just like any other.
We have records of both.
And we don't have nearly the same amount of detail in women's work. That's why even today it's referred to as invisible labor.
[citation needed]
>Except that kind of work would have been split up by gender.
It's still work. It doesn't magically have zero value because it was done by women. The methodology cited by the book counts only counts hours worked if it's farming or wage employment, but doesn't count any household labor, which means anything that's shifted from being made at home to being bought would could as extra hours worked, even if it saves time overall.
Nobody said it had zero value. The point is that labor roles were extremely gender striated, and the record is mostly kept only for men's labor. So saying that they worked more because of the example task you gave doesn't make sense when societally that work was put onto others.
Tractors are tools to make hard boring labor easier.
Even after the mechanical loom was created, figuring out the next most important problem to work on was a job that humans did.
Unless you believe there's some hard limit on AI intelligence that will constrain it below the intelligence of a particular earthbound hairless ape, then eventually AI will be perfectly sufficient and probably better at figuring out the next most important problem to work on.
Ta-da, humans are completely removed from the value chain. Neither the loom nor the tractor could not do such a thing.
The biggest issue we face today is the incessant nature of economists to try to reshape reality to match economic theory, and not the other way around
You have it totally backwards lol. You can remove the economists or even the word "economy" and nothing would change about actual reality.
The fear is this will replace engineers, scientists, accountants, lawyers, service workers, etc, etc all within a small time window.
Investment types have shown repeatedly that their primary concern is money, not workers. There is no reason to believe that those currently in power are open to sharing their wealth or influence.
If you stayed on the land you had to work, not quite like a slave, but close. And if you disagreed with this, the government had an army that convinced you ...
Factories offered a better alternative than that, and yes, mostly because the agricultural option was just not open, and just not worth it. They also offered a great density of people that made the labor movement possible in the first place.
Most of the writing comparing factory work to agricultural work is from after the enclosure movement forced large portions of the agricultural working class to take the "wage work" arrangement with a similarly brutal schedule and work pace to factory work. You're right that the technological and agricultural advances that led to enclosure created a source of poor workers willing to accept factory work.
Obviously the peasants weren't spending the rest of the time sitting around, they had to work on their own land to keep themselves fed and clothed. It definitely wasn't some sort of lazy paradise, subsistence farming is a brutal and unstable way of life, I'm just arguing that the early Industrial period was even worse. If you want an "objective" measure of quality of life across vastly different time periods, the height of the average European went down a few inches during the Industrial Revolution due to malnutrition and overwork, and only recovered to its preindustrial average in the first half of the 20th century.
Now with the tech era, you can snoop into people businesses and control in a "softer" way. With AI, you might be able to bypass the knowledge workers who enjoyed a pleasurably life in the last 30 years or so. This is mostly because knowledge workers will sabotage your operation if not treated well.
Now this might not longer be the case as you'd no longer need them. The revolution can't happen if the people have no leverage: 1. either be able to defend against a professional army or 2. are an input in your supply chain and can stop your operations.
Who's definition are we talking about here
But it unambiguously is who is leading the promise to replace labour; it is whose bullshit is provoking the backlash.
It is whose variously-confident definition of near-term AGI involves completely eliminating a large enough shar of jobs that Altman thinks he has to talk about UBI and fund pilot studies so he doesn't lose all his cool California friends.
It is who has set the tone for the entire punditsphere and everyone who emulates them.
Their definition is the one the media and governments have swallowed; they need the fear so they get regulatory capture.
I will tell you that I think though: I think Sam Altman should be nowhere near the power and proximity to power that he has, band I think Dario Amodei would be better off tucked away in a research unit where the only person who has to listen to him tolkien bollocks is his immediate boss.
I think they are manipulating truly grotesque amounts of fear that are in many ways worse than the fear we felt in the COVID pandemic, and I think they are doing it for money and power.
But I also think people have listened and the message has not got through to tech people that they aren't good people and they are fucking weird.
And the farm analogy is somewhat on point. We went from 67% of people working on farms to... I think it's more like 3% than 1%, but a very small amount. That's two thirds of labor being replaced.
It's so common here and so obviously wobbly. That labour was displaced (and in most cases into way more gruelling and dangerous factory work).
The AI pitch is that the giant superbrain will do all the knowledge work and rapidly self-improve faster than humans (and therefore, do more future jobs we could do). That is a pitch for replacing human labour.
You can't simultaneously have a machine that is said to be likely to wipe out entire categories — not market sectors, categories of work — and then say that all those people will get jobs elsewhere. Because, where? The timescales they are talking about are short. Where's the work going to come from in time?
The next couple of years will see these people contorting themselves into increasingly complex knots to try and argue that AI is making the world a better place. The arguments will get far weirder, and far more detached from reality, than bad comparisons to agricultural labor.
I mean, if it delivers. So far we're only really f**ing ourselves in the face; outside the tech industry everyone else is figuring out how to push back on AI.
If you're never seen Man of Aran, you're in for a treat! Far more thought provoking than the fluff Lucas made, but by which I always assumed his fictional homestead on Tatooine was inspired.
Similarly with LLMs, it's good tech, but I do not believe for a single second that these things will ever replace anyone who is remotely skilled. People will be augmented by them. They won't be fully replaced. Well, I hope not anyway.
Surgeons already perform remote surgery, so physical dexterity isn't the issue.
What do you think the moat is? Computer vision? Surgical technique? Surgical knowledge? Medical knowledge?
Maybe surgeons will be smart enough not to allow AI to be trained to do what they do, but it seems that training data is the only real barrier.
https://www.news.com.au/sport/carnage-at-start-of-robot-mara...
We might have to get off the computer, and we might have to rethink how we organize the world economically, but there is still work to be done everywhere.
This is an absolutely bizarre pitch for labour replacement for the very obvious reason that the rate at which the world's hair grows is not going to increase and nor are we going to suddenly discover a great need for new high maintenance hair styles to increase the work available.
There are clearly at a first approximation enough barbers and hairdressers already to cut the world's hair at the rate it is growing.
Essentially every single medical school in the USA is vastly oversubscribed with qualified applicants, as far as I am aware. And most specialist residencies are too.
But you know you can't just become a doctor, right? You don't just lose your job and become a doctor, or a trained nurse the next day. That is a ten, twelve year lead time thing for doctors, two or three for a nurse, and it favours the young. People who lose their jobs mid career essentially never become doctors. Some do become nurses, but it still favours the young.
Plus, even if there is a shortage beyond that, and in general pratice there is, we don't need hundreds of millions more doctors around the world, I suspect, but that is how many jobs AI threatens to displace according to the FUD. We maybe need a few tens of millions more nurses.
And if you expand the healthcare and training system to train up and employ more people — where does the money for that come from?
People who simultaneously go along with the idea AGI will eliminate large swathes of human labour but somehow magically think everyone will find a new job so there is no retail and service demand collapse or tax revenue collapse are everywhere, serving up the tractor analogy, which is where we started in this thread. It's horseshit.
The 90% of workers in agriculture found jobs in industry and services.
Those industry and services will soon largely be performed by AI, with the exception of the lowest paying jobs that are hard or expensive to replace with robots (construction, plumbing, hair styling) or for which demand for actual humans is a key ingredient (prostitution, live entertainment).
So the 90% of workers in today's industry and services will find jobs in ... (fill in the blank)
The problem is, I have not read a single reasonably logical prediction of what _blank_ might be.
I reject this argument as bad faith from the start.
AI is an interesting experiment with some real-world applications, but it’s no tractor or mechanical loom... not yet, at least, and it’s far from clear that it ever will be.
As things stand today, AI is not the future. It’s a tool with genuine uses that is being marketed as a revolution.
And how many years did that take to happen? 3 or 4? That's what the AI companies are promising. 89% of you will be unemployed in the next couple years. All that wealth you'd be making from working will now be going to the company owners and you're out on your own. Good luck!
The industrial revolution came along with massive production of goods that people need and desire. Even then, there was still a huge amount of pushback (it still echoes in a lot of communities today!).
Do you see any differences compared to the AI revolution we're being sold?
Do you not think the ultimate outcome was worth it?
But if you're not a troll, from the very fucking start of it there was a boom of products. Like it's practically the definition of it. Are you seriously thinking that they just burned a bunch of coal for fun?
I never claimed it.
But if we play along, and if I claim AI is equivalent, then once again ... why ask about glut of food and consumer products only a few years in, when that wasn't the case with the Industrial Revolution?
> But if you're not a troll, from the very fucking start of it there was a boom of products.
It began in 1760 in England. Can you compare 1764 with 1760 and itemize this "boom" you speak of?
The changes the Industrial Revolution brought about was slow - it took several decades. Correspondingly, the change in labor took that long as well.
2. If 1 happens, it is foreseeable that they don't even need that many slaves to tend to the machines, so not many "new" jobs created I'm afraid.
Comparing that to the computer -- yes if you are a typist you are doomed -- but they still need someone to type -- just not on a typewriter. And there are suddenly countless new requirements (e.g. video games, CGI, etc.) getting created from thin air. I don't really see this happening for AI -- like, do you see any NEW requirements getting created? Sure we are getting endless AI slop games/videos/fictions, but are they new? People can only consume that many products and pumping 100x into the system doesn't work -- except to make profit drop to ZERO for every ordinary creator out there.
BTW I do think there will be new requirements -- robotics combined with AI (e.g. who doesn't want a handsome husband or a beautiful wife?) -- but again it is to replace humans.
3. Apparently, accountability has dropped to a new low, since the end of the Cold War. So naturally cattle and sheep (I mean, us, ordinary people) are scared. Back then at least the elites were willing to make gestures and put up facades. Nowadays they simply don't GAF.
Look at your c-suite executive team and you'll find it stacked with them.