I wonder why people reflexively quote a "fallacy" from the year 1891! It wasn't true even back then, as evidenced by the multiple economic crises that followed.

Nowadays we have a completely different economy to begin with, saturated with "bullshit jobs" (Graeber) already.

The amount of new inventions is finite. In the 1990s we had TGV, Concorde and Maglev trains. Perhaps physical inventions have been somewhat exhausted?

What invention is on the horizon that will provide new jobs for those laid off from "bullshit jobs"?

How did society support Einstein when he discovered relativity? It didn't: Despite the invention of the tractor in 1892, which, according to HN commentators, should have provided him with a carefree life, he had to take a job in the patent office. Which, according to "AI" fanatics, would now be automated by "AI".

I always get a little triggered when i hear the phrase job creation. it’s like a solution in search of a problem. here idk if you’re using it in the way i usually hear it because i generally agree -productivity gains boost the economy. for me, i wish the focus was on value creation. i think of the economy like a bike, the slower it goes the harder it is to balance. i come from a biology background and there there is ‘boom bust’ as a widespread and normal process. if humans are clever enough, even when we run out of resources and are about to bust, we can just suddenly invent fertilizer or countless other things. we took natural gas and added value to it. humans have lots of tricks to keep the bicycle chooching along. Regarding ai, it’s a systemic change. most people aren’t great at systems thinking because most people are specialists. so it’s not too likely you’ll hear a salient take on what ai will do despite some rather smart people commenting. The one trend i like to look at is sort of a scale reset or, how do you say forest from the trees? what i mean is, one of corporate’s great advantages are their size and swath of roles and the coordination that allows them to have an outsized advantage-essentially leveraging the collection of specialist to gain a dominate effect in working ‘the system’(navigating government, economy or scale, overwhelming capital, etc). Enter AI, subsuming these roles into one thing and sort of resetting the required scale to have some of the corporate power/advantages. Of course these are some rosy shades but i often approach new things with “what’s the best that could happen” and it has served me well.

They boost the economy if there exist enough mechanisms that spread the generated gains to everyone. Those mechanisms used to be unions, political power of the working class and the demand for labor. The first two are close to nonexistant now, and the third one is being eroded at an increasing speed. The closer to human abilities automation gets, the less room there is for new human jobs to be created. At some point any new job invented can also be done by an AI, and at that point productivity gains have no mechanism with which to spread into the economy. At that point the owners of the machines will get everything they want from them and they'll have no need to pay other humans anything.

Economics is a pseudo-science typically used to launder political ideology, but this "fallacy" is particularly ill suited to this discussion. Yes, it's true that automation doesn't necessarily cause a net unemployment across the board to go up. But it can still decimate employment in the automated sectors. From that link:

>While many workers fear that automation or artificial intelligence will take their jobs, history has shown that when jobs in some sectors disappear, jobs in new sectors are created. One example is the United States, where a century of increasing productivity and technological improvements changed the percentage of Americans employed in the production of food from 41% of the workforce in 1900 to 2% in 2000.

The problem with AI is that it tends to automate away skilled jobs, ones that sometimes required many years of study and educational debt to get.

So the net result, with respect to software engineering, which was the context of the discussion is here, is not "job creation." The net result is that people (especially the junior ones who just got out of school with a pile of debt) are now forced to compete for jobs they aren't necessary any more qualified for than people without those degrees.

This applies to other things where AI massively reduces the number of people needed. Journalism, art asset creation, etc. These are the kinds of cushy and often fulfilling jobs that are only possible because our grandparents and their parents, and so on, worked backbreaking or mind numbing jobs to build the kind of economy that would support these kind of careers. Thanks AI! My kid might never have to worry about the horrors of creative engineering or artistic careers, freeing up room for the real joy of being a gig economy slave or a factory shift worker!

Literally trickle-down economics.

Yea, I believed that for a while and still do...but we're going to go through a rough patch first. There's a bunch of people who don't understand how this works and are looking for a get rich quick scheme. Very lazy people. They're going to further devalue employees and expect unrealistic outcomes.

It's going to take some time here before people learn how AI fits into the every day and along the way there's going to be some wild stuff.

I also expect there to be a very very strong addiction or cult-like behavior. Mental health will suffer as well. Very dangerous times.

To be fair the firm's goal is, explicitly, to "replace most economically valuable labor".

"The economy" (GDP) being boosted is very unevenly distributed.

Are the jobs created good jobs - that pay decently, and that people want to do?

uncontrolled productivity gains that don't feed the added value of that extra productivity back into the system is failed. you end up with the 0.1% having all the money.

hmmm... wonder what country has this balance.

Job creation for robots, yes. Remember that now workers will come in two flavors, human and robot.

But robots need no compensation, no health insurance, they don't have any rights, can work 24/7, they cannot sue the company because they have no legal representation, cannot steal from the company, they can suffer an "fatal" accident with minimal liability to the company, they can erase their memory of any liability to the company, will always say yes to everything, training them (for a human = pre-k, kindergarden, elementary school, middle school, high school, college, graduate school, etc) for them is transferring a file or receiving commands wirelessly from a central computer, they don't get bored and quit, they don't get unhappy and start looking for jobs. Good luck competing against that.

Once one ML model automates a job, you can make a million copies of that model file and that job is gone. Similar to the Borg (one adapts, all adapt). Once robots start building robots, you get an exponential amount of robots that quickly take over.

A human needs protection from temperature, noise, injuries, exposure to chemicals, exposure to biohazard, protection from other workers, from management, from animals, from their work equipment and gear, and... from robots! Every aspect of a human is protected including their privacy, biometrics, etc. On the other hand, robots require protection from: nothing. They are completely expendable and their cost will be driven down over time.

If you make a robot 33% as productive as a human per unit of time, because it can work 24 hrs instead of 8 hrs, it has the same output.

Society in 2025 is a pyramid scheme based on scarcity and now the base of the pyramid will be taken over by robots. The people who are left outside the system will be the government's problem. But unlike in the industrial age, large amounts of people will no longer be a key to power so their political representation and rights will be evaporate. There will be no universal basic income (that would result in humans multiplying geometrically and eating up all the resources). Robots will be the key to power therefore who controls the robots will control the government, that means, oligarchs or superhuman AI. And then at least one oligarch will give full autonomy to the AI to prevail against the other oligarchs (humans are bottlenecks), and in doing so the one who rules everything will be superhuman AI.

You may think that people in the future will have cheap access to AGI and therefore be able to scrape a living somehow... the problem is that once large AI conglomerates get all the data they need, access to cheap AI will be cutoff and all the deskilled masses will be left with will be endless brainrot social media feeds and nothing to eat.

Productivity gains also boost wealth inequality. Too much wealth inequality has negative effects on democracy and society in general. AI has the possibility of making the tech billionaires far more wealthy and influential.

This is true under an economic system that appropriately distributes the fruits of economic production to workers; perhaps one that existed in the post-ww2 to neoliberal era. But it is untrue under a rentier / technofeudal / oligarch economy, in which all excess profits are captured by a cabal of the owner class. Which of these states describes the modern economy is a political exercise.