Hacker News Reader

Firstly it's not even clear that AI will eliminate jobs without creating new ones.

I would say that's pretty clear in the long run. AI will most likely create jobs, but how many jobs can it create ? Every white collar job potentially affected against what exactly ?

but you gotta cross that bridge when you get there, if you get there

So we can have more of the "gene out of the bottle" or "train has left the station" response ?

How many jobs can those damned steam engines create anyway?

And how good of an analogy is that ? On one hand there's something affecting a part of transportation (of course transportation directly or indirectly affects other things) and on the other something that will potentially affect every output of the human mind ?

> On one hand there's something affecting a part of transportation

Yeah, it's not like there was an industrial revolution or anything...

Ok my bad. But still, why is it a good analogy again ? How does it compare to where we are. Walk me through it.

> AI will most likely create jobs, but how many jobs can it create ? Please walk me through the parallelism with the Industrial Revolution.

The impact of AI on jobs is a topic of ongoing debate among experts. While some argue that AI will lead to significant job losses, others believe that it will create new opportunities and stimulate economic growth.

To understand the potential impact of AI on jobs, we can draw parallels with the Industrial Revolution. The Industrial Revolution, which began in the late 18th century, marked a period of rapid technological advancement and social change, as new machines and manufacturing processes were introduced.

Initially, the Industrial Revolution led to job losses in traditional industries, such as agriculture and textiles, as machines replaced manual labor. However, it also created new jobs in manufacturing, engineering, and other industries. Over time, as the benefits of industrialization became clear, the economy grew, and new opportunities emerged.

Similarly, AI is expected to create new jobs in industries such as robotics, data analysis, and software development. These jobs will require specialized skills and training, and may not be accessible to all workers. However, they are likely to pay well and offer opportunities for career advancement.

At the same time, AI may also lead to job losses in certain industries, particularly those that rely on routine, repetitive tasks. For example, automated manufacturing systems may replace human workers in factories, and self-driving vehicles may reduce the demand for truck drivers.

Overall, the impact of AI on jobs is difficult to predict, as it will depend on a variety of factors, such as the rate of technological progress, the availability of training and education programs, and government policies. However, history has shown us that technological advancement can both create and destroy jobs, and that the economy can adapt and evolve over time.

Thanks chat gpt. That's a very 2021 answer - definitely what we thought up till then. But the most curious thing is:

both have led to job losses due to automation, but also created new jobs for skilled workers who were needed to operate the machines or run AI-powered systems respectively

which does prove my point, no ? Operate machines across all industries vs data science ?

>> created new jobs for skilled workers who were needed to operate the machines or run AI-powered systems respectively

> Operate machines across all industries vs data science ?

Operate the machines across all industries vs run AI-powered systems across all industries.

(Those chatgpt-like responses miss a major factor, though. Automatisation of agricultural processes didn't result in 95% of the population operating agricultural machines: it freed them to do other things.)

run AI-powered systems across all industries

Sooo...data scientists ? Isn't the focal point of all of this the realization that given time (most likely a lot less than people thought before gpt's) AI systems won't need human oversight ? Isn't this the goal of OpenAI ? This is why I think making comparison to whatever that was is wrong. Our current situation is too distinctive. We're being potentially freed up to do things other than cognitive tasks - by the rise of AGI, or maybe a very refined language model.

>> run AI-powered systems across all industries

> Sooo...data scientists ?

If someone using an AI-powered system is a data scientist (?) does that make someone using an electricity-powered system an electrical engineer?

I don't know. It's out of my scope that's why I'm asking. The infrastructure is there, the software is there and if not it soon most likely will be. What's left is to feed it domain specific data. To contribute somehow to a system like that I imagine you have to have a cognitive ability way beyond average. All the 'AI-' in front of job descriptions won't be in the ballpark of greasing cogs during the industrial revolution.

The previous response courtesy of chat.openai.com - I tried the same prompt using gpt4all on my laptop and got the following in less than 20 seconds. Really impressive!

-----

> AI will most likely create jobs, but how many jobs can it create ? Please walk me through the parallelism with the Industrial Revolution.

The creation of AI-powered machines and systems is expected to generate new types of jobs in addition to replacing some existing ones like data entry clerks or accountants. However, there are concerns about how many net job gains will occur as a result of the automation revolution that has already begun with robots taking over repetitive tasks such as manufacturing assembly lines and warehouses.

The Industrial Revolution brought significant changes to society through mechanization in industries like textiles or steel, but it also created new jobs for skilled workers who were needed to operate the machines. Similarly, AI will create opportunities for professionals with expertise such as data scientists and machine learning engineers that can help design algorithms and train models using large datasets.

The parallelism between these two revolutions is evident in their impact on society's labor market: both have led to job losses due to automation, but also created new jobs for skilled workers who were needed to operate the machines or run AI-powered systems respectively.

If your job gets taken by an AGI and that creates a new job that it can't do, it's not an AGI. The only things a human-level intelligence can't do but a human can are physical labor (provided that robotics are not available). That is distinctly different from the industrial revolution.

And with an AGI I am certain we can have a robot race building more robots autonomously, thus a reproducing organism, who like all others at the top are hell bent on maximum

Steam engines and their descendants largely replaced the jobs which were done by horses and mules, and they did that successfully - there are some jobs for horses still remaining, and initially many new jobs were created for horses in cooperation with the engines (peak-horse was at 1920-0ish), but the total number of jobs for horses has collapsed to nearly zero compared to what it used to be.

So the analogy of steam engines suggests that this can be quite dangerous indeed.

Also, we do have historical evidence at how fast human society reacts to changes when technology causes great shifts in needed jobs - not fast at all. Even getting to terms with decades-long changes caused significant social upheaval and strife, as in, for example, the automation of textile industry; If the current speed of technological change would cause a disruption in jobs which happens over the course of just a few years, it will be hectic and violent even if there is the same number of new, different jobs created.

> Steam engines and their descendants largely replaced the jobs which were done by horses and mules

> for example, the automation of textile industry

How many horses and mules were working in the textile industry again?

As you have noticed, the last paragraph is about other types of automation during the industrial age than the steam engine.

What do you think that was powering that automation exactly?

In the textile industry many of the major changes were from better devices, not neccesarily of how they were powered.

The cotton gin revolutionized the industry while it was still people-powered, including by slave labor (edit: and also horse-powered); and the Luddites (some decades after Ned Ludd) rebelled not against the looms being powered by steam but rather against the Jacquard loom - an add-on to the exiting looms which allowed complex products to be weaved by unskilled people, removing their bargaining power despite not changing a thing about how the loom was powered.

The "how they were powered" was quite important in the magnitude of many of those changes though.

"Although some were driven by animal power, most early mills were built in rural areas at fast-flowing rivers and streams using water wheels for power.[3] The development of viable steam engines by Boulton and Watt from 1781 led to the growth of larger, steam-powered mills allowing them to be concentrated in urban mill towns, like Manchester, which with neighbouring Salford had more than 50 mills by 1802.[4]"

https://en.wikipedia.org/wiki/Cotton_mill

Or worse some silly French commanders throw legions of frogs over the dugout, this time not against mechanized machine guns, but literal sentient death machines. How high the toll would be

> So we can have more of the "gene out of the bottle" or "train has left the station" response ?

Your concern is theoretical "social unrest" due to being replaced by automation even in the face of factual social unrest due to extension of retirement age.

Even if automation replaces jobs en masse, It's not as if this would happen in a vacuum even if it does. We already have social safety nets (even in US) like unemployment insurance, medicaid, food stamps, etc not to mention a robust private charity network. And government already showed, for example, during the pandemic, that it was willing to fund people in an emergency if and when it happens.

One thing I do know about AGI/AI and automation in general, is that it's not going to sweep through entire industries overnight all at once. It takes time to deploy, time to deal with edge cases, time to refine, time to reconstruct processes around automation, time to adapt to whatever downstream changes and efficiencies the automation creates, time to effect prices in the marketplace, etc. Time to see if other jobs not previously anticipated when a new tech replaces an old one, like "web designer" or "web developer" or "SEO specialist" did with respect to newspapers or letter carriers or television programming being replaced or reduced by the existence of the internet.

Those safety nets are not enough to live on for a family.

Are we really worried about elderly revolting?

>At any rate, if AI is busy automating away labor, then the world is getting more efficient. And that means there are more resources to support people.

If the world is getting more efficient, then why would those resources go to people that are intrinsically less efficient? The current state of the economy is to funnel capital to those who efficiently allocate; the amount of intentional inefficient allocation reduces as efficiency grows (as the world gets more productive, less of this output lands in the hands of workers). Why think that the distribution model will change significantly in a timescale relevant to people?

Our culture takes it as axiomatic that more efficiency is good. But its not clear to me that it is. The principle goal of society should be the betterment of the lives of people. Yes, efficiency has historically been a driver of widespread prosperity, but it's not obvious that there isn't a local maximum past which increased efficiency harms the average person. Historically, efficiency was driven by innovation that brings a decrease in the costs of transactions. This saw an explosion of the space of viable economic activity and with it an explosion of prosperity. Productivity and wages remained coupled up until recent decades. Modern automation has seen productivity and wages begin to decouple. Decoupling will only accelerate due to AI. We may already be on the other side of the critical point.

This is an excellent analysis, but I suggest that the state of the average person is not yet at a maximum. There are still thousands of major, valuable problems to solve. The efficiency offered by new automation should enable more people to become more powerful problem solvers.

Counterpoint - what if AI automates away that labour and _nothing_ replaces it for the people who are automated away? Our history has not been kind to people whose jobs have been obsoleted, why would we assume any different this time, that because the technology that is obsoleting jobs is AI that the beneficiaries of that enhanced efficiency would choose to spread those savings back around? If they cared about that in the first place, why would they even chase the efficiency of replacing people with AI?

I think this is a bridge you have to strongly consider if we want to cross _well in advance,_ given that it is likely to set on fire and fall down behind us if we do, trapping us on the other side whether we like it or not.

It's not replacing people, it is doing something else than we do. Even the fact that AI is scalable and doesn't get tired indicates the job replacement idea is wrong. AI will do massive work, something we couldn't have done with 100x more people.

Our mistake is to be human-centric. We think our work will be stolen by AI bots, but we can't imagine AI creating business and hiring people. When AGI comes, we will have to care for the needs of humans + the needs of AGI. And it has needs, so there will be jobs. And AGI might be more prolific in development than we are, there might be plenty of work for us.

It's worse than that. You refer to the mythical "we" that is able to act coherently and make effective decisions about AI use. There is no such thing. The world is a set of competing individuals, countries, and corporations; each with its own interests, acting independently.

I don't think many people see this because the "mythical We" is used all the time. "If "we do this", "we must do that", as if all of humanity can act together coherently or ever has. People seldom think about how imaginary this viewpoint is.

Exactly. There's no "we", just competing interests.

You're 100% correct, I just didn't want to sound _too_ doom-and-gloom about AI, it feels like that's a quick way to get massively downvoted.

I'll just add that I have so much more "non economically viable" work than I'll ever be able to do. I'd love machines that could write code for me to do tasks that I'll never be able to afford to invest in otherwise.

put another way, we have a labor shortage at today's level of output already. BUT, I don't know anyone who doesn't want more output. I mean, we're not going to run out of stuff to do!? what about space? curing diseases, etc. what about having time to make more art?

if an AI can do work, or make work easier, what makes us certain that we'll just settle for getting the same amount done vs doing more?

>if an AI can do work, or make work easier, what makes us certain that we'll just settle for getting the same amount done vs doing more?

In this hyper-efficient future, is being paid for this stuff going to be viable? Why would anyone allocate money to your inefficient endeavors? There's this strange blindspot in people who think that more economic efficiency means you'll have time to do intrinsically inefficient labor. Who is going to feed and house you when your labor is irrelevant to the workings of the economy? UBI is a nice dream, but how do we get there from the current ruthless efficiently allocating system?

One of the proposals out there (often posed as part of funding UBI - but not always) is to tax robots.

Should we tax robots? - https://news.mit.edu/2022/robot-tax-income-inequality-1221

> What if the U.S. placed a tax on robots? The concept has been publicly discussed by policy analysts, scholars, and Bill Gates (who favors the notion). Because robots can replace jobs, the idea goes, a stiff tax on them would give firms incentive to help retain workers, while also compensating for a dropoff in payroll taxes when robots are used. Thus far, South Korea has reduced incentives for firms to deploy robots; European Union policymakers, on the other hand, considered a robot tax but did not enact it.

https://en.wikipedia.org/wiki/Robot_tax

There is a lot of material out there both for and against taxing robots.

No, i mean, I don't expect anyone to allocate anything to my non-viable projects. I expect to allocate my own money to them like I always do, but, I'll be more able to achieve my goals. If things pan out, maybe these things could end up as a bootstrapped startup.

Efficiency here could mean less need for equity funding and angel investing in early stage projects.

I think the issue is how fast it could happen. What if 75% of most white collar jobs disappear in 2-5 years. What would happen to the economy and demand? And let's assume the 25% left is out side of skill set to be retrained within the month of training most corps are willing to spend training someone.

That's the issue. We don't know what would happen but if what you said were to happen, governments will ban AI since the current economic model (capitalism and taxation) will be render useless and society itself will demand it (AI ban).

What if a different statistically just as likely utopian scenario happens instead and we generate more jobs than we can keep up with because we start colonizing star systems as a result of using AI tech to discover some new propulsion technology? What if AI can't do anything useful without a human in the loop to guide, monitor, assist? What if groups of humans augmented with AIs are far more powerful than AIs alone? What if the implications of that are that we create advancements in biotech, energy efficiency, software and information systems, interactive entertainment, agriculture, production, mineral extraction, etc that cause an explosion in jobs?

>It's not clear that any development in AI will be any different than the transition to mechanization in the 19th and early 20th century.

so you mean utterly hellish?

>https://martyrmade.substack.com/p/martyrmade-22-whose-americ...

Coal miners protesting drastic conditions in dark shafts in raw mineral extraction mines in 1921 doesn't tell you anything about mechanization and the effect it had on people's lives, not to mention electrification. It wasn't possible to mechanize coal extraction until later and even today with superior tooling and safety conditions, coal extraction remains dangerous.

In the meantime, for vast majority, it improved standards of living, hourly wages, etc. You can't just look at arbitrary data points in early 1900s or late 1800s. Hellish wars, famines, civil strifes, rebellions, assassinations, religious battles, etc go back millenia.

> https://www.mlive.com/news/detroit/2018/01/historic_photos_f... > https://www.history.com/this-day-in-history/ford-factory-wor... > https://worldsfairchicago1893.com/ >