Hacker News Reader

Comment by lgieron | Hacker News Reader

> I could make the same argument about farm equipment eliminating farmers' jobs, and not leaving as many jobs for farm equipment repair.

This. 300 years ago, 90+% of population was working in agriculture - now it's around 2% in developed countries. And yet, there was no armageddon - we managed to develop all these new needs that keep people employed. People saying that "this time it will be different" ignore the almost unlimited growth potential of some industries, such as for example teaching (we could go back to the standard of one teacher per one pupil that rich people had back in the day), psychological counsel (potentially everybody could benefit from a shrink/coach), or human-based health care (more nurses -> better standard of care).

View on HN

That all sounds lovely. Except modern automation means further concentration of wealth in the hands of a microfraction of society, leading to not being able to pay for any of that because a monopsonistic economy means very bounded demand which leads to crushingly low wages. And any sort of skilled-service jobs require an economic base to get into in the first place--money that you don't have because the money is not in your or your parents' hands. BRB, reinventing peasantry. Or worse.

Economic activity doesn't magically happen when all of the lubrication for the engine is left in a bottle on the shelf, which is what modern automation is happily barreling towards barring a significant reworking of society that will have all the invisible hand fiction-peddlers in the tizziest of states.

I'm not sure this is wrong, but I think it's worth recognizing the two major objections.

1. Capital ownership and opportunity: All of the careers you mention could absorb many more workers in terms of available tasks (something like teaching could easy grow 10x), but that doesn't justify a claim that they will. The farming transition was enabled by mass ownership of productivity gains - careers like entertainment expanded massively because lots of demand was newly available to support that supply. The Picketty school of thought argues that this isn't inevitable; it's possible that a few kids get one on one education, and a bunch of people starve because wealth concentration means there aren't enough buyers for their work.

More specifically, the rich won't buy arbitrary amounts of services because their time and desires are not unbounded. Even 1-on-1 teaching can only employ one teacher-hour per buyer-hour, and the returns on 'background support' don't really justify having 50 people make one lesson plan at living wages.

2. Capability: Take a look at all of the careers you outlined there. They're all skilled-service work, most of which requires having more of some trait than your customer. Teachers need knowledge, counselors need insight, care workers need knowledge or patience and emotional reserves. Crucially, a lot of this is relative - it's not enough for a teacher to know things, they have to have knowledge their customer doesn't have and does desire. Factory labor took some specialized knowledge, but the transition from "a strong back for working fields" to "a strong back for milling cotton" was far less extreme.

This is the sort of two-sided market that's hard to stumble in to. I'm not saying most people can't learn an employable social/service skill, but if you lose your factory job you can't just accept a low wage to work as a teacher. There are significant time and price barriers in that transition, and it's not clear that they'll be overcome without non-market intervention. Markets are wonderfully efficient, but they can certainly decide that it's not worth retraining someone instead of letting them starve.

To further detail the parents point on Capability and "nature of automation";

Automating processes in production activities aims to "remove" human factor(s) altogether from phyiscal activities, leaving only management and maintence personel (that is if it is not outsourced) of the automated systems in a production site. Thus, comparison between "reduction of workforce" because of more advanced tools (and animals?) and "removal of workforce" caused by automation is just... silly.

There are many reasons to automate processes, chiefly eleminating or reducing certain costs and increasing quality "consistency", as well as decreasing activity times with all the other benefits. The latter of which may be the most important for certain sectors.

Even then, if the production site managers deem it is necessary to keep workers to preform supporting activities (eg. keep an eye on dem robbits leaking lubricants), it will be either because they are still in process of being automated or quality control needs a human to "feel it" or its just a publicity stunt to show they are fullfilling their obligations to society (ha!). But automation will ultimately result in "no jobs" for workers who did the activities that became automated; there is absolutely no reason to keep workers, even with lowered wages. Remaining human workers will have to have higher qualifications and move on to a different set of responsibilities, eg. as mentioned before like maintance of automated systems; inspection, servicing... etc. instead of... y'know, doing what the automated systems do. This will effectively eleminate the need to employ low-to-unqualified workforce in a production site. Even with all the best intentions to employ maximum number of workers, total number in a production site, after fully automating production processes, will be a fraction of what it was before automation.

note: written while drunk, may not exactly be english or even engrish.

We have to move ownership of automation in the hands of the population, in order for it to become self supporting / self sufficient. We used to be self sufficient when most of us lived on farms. We can do it again, at much higher densities, with automation.

I think one of the key differences this time is the amount of retraining involved. Taking a farm worker and teaching them how to work on an assembly line involved orders of magnitude less retraining then training an assembly line worker to be a teacher, psychologist or nurse.

Also if you're into human biological differences and heretical things like IQ, the bell curve bar for "can farm instead of hunter gather" "can factory instead of farm labor" is very low, like IQ 70 or something. You don't need an IQ of 120 to put a nut on a bolt on an assembly line.

However, lets say all the jobs with IQ below 120 are replaced with an IQ 120 AI or similar automation via the efforts of over IQ 120 engineers...

Most of the population is physically unable, as best we understand, to ever obtain a MSEE or a PHD in computational fluid dynamics. It doesn't matter if we have 100M job openings for people with advanced CS degrees no matter how much we pay them, with only 300M people in the country the bell curve says we can't find the warm bodies to staff them.

Applying machine learning is not going to be requiring a PHD, especially when great libraries and model zoos are available off the shelf. I think it's going to turn into a "maker culture" like web design. People are going to enjoy a very low entry barrier to applying ML and automation. It's going to be mix and match, a bazaar of solutions. Most of ML will be assisted by meta-ML systems and thus almost automated.

> 300 years ago

What if this dramatic change happened over a span of two-three decades?

Reply on news.ycombinator.com