70 Comments
User's avatar
RenOS's avatar

It's insane how much we're arguably already living in the world that AI pessimists fear, i.e. only a minority of people are involved in important productive work that we actually need, and an even smaller percentage are involved in important research and development that has a realistic shot at improving our society.

Many people already waste substantial percentages of their life on unproductive endeavours, and/or choose service sector jobs that are as much a service to themselves as to others. Streamers or game developers are of course the purest examples, but many coffee shop owners, political activists, or almost anything relating to pets, particularly expensive ones like horses, and of course sports, as you mentioned. Anyone who has interacted with it knows that the bureaucracy is partially a place to generate make-work for people who wouldn't cut it anywhere else, and another part is actively inhibiting productive work of others. And most people I know who work in large companies will have plenty of stories of dysfunction, bloat and inefficiency, even if the company might at least in principle do something worthwhile.

If the percentage of work humans perform which is objectively important goes from 10% to 1%, would the average person even notice? 1% to 0.1%? Does it matter to their day-to-day? I don't think so. And as you say, entertainment, make-work and status games can be indefinitely generated for the rest.

Almost everyone is simultaneously too cynical and not even nearly cynical enough.

My bigger worry is the possibility of disenfranchisement if the AI really is that much better than us, becomes fully agentic and holds the majority of the levers of power, but there are quite a few more steps from here to there.

Richard Hanania's avatar

Excellent comment. The pessimists fear making it possible for everyone to have jobs that are as pleasant as those of people with the most enjoyable jobs today like artists, actors, writers, etc.

Scott Sumner's avatar

Excellent post. This is interesting to think about:

"He was only wrong in assuming that the work week would be shortened, when in reality we simply extended the periods of life during which humans don’t engage in the labor force."

People often make fun of Keynes's prediction, but (as you suggest) much of our current work is actually a disguised form of play. If you subtract all the time office workers spend surfing the web and chatting with coworkers, I wonder if we don't already have the very short work week that Keynes predicted.

Anonymous's avatar

The reason that many pessimists fear that the gains from AI will not be widely shared rests on the assumption (a correct one in my opinion) that AI enables BOTH job automation and systematic political repression through an advanced surveillance state. Past technology advances lead to a rise in social welfare spending because of a credible societal threat of armed revolution in its absence (like the infamous 1932 veterans march on Washington DC). It was cheaper to "pay off" retirees and the dusplaced than to quash them. This revolution threat was credible because of military drafts that resulted in a high percentage of the civilian population gaining military tactical skills.

In an AI-tech world a few oligarchs can counter any potential revolt with automated drones and robots that only they control. With the outside bargaining option of revolution off the table the AI oligarchs then lack any economic incentive to tolerate a tax-and-spend welfare stare.

Richard Hanania's avatar

It's extremely cheap to massacre citizens, or kill enough to scare off the rest. Government can do it easily right now and save on their budgets.

They don't because of morals and ethics and people getting better over time. People care about virtue, or at least appearing virtuous. Part of the historical process is that as you get richer, you can afford to be generous. For billionaires today, it objectively doesn't matter that much if they have $1 billion or $8 billion, while a Southern plantation owner losing his slaves would be disastrous for his family. So people with greater wealth become more generous.

This will be much more true in a world where AI supercharges productivity to previously unimaginable levels.

Dean Valentine's avatar

> Government can do it easily right now

Complete nonsense. The fact that government officials don't know if soldiers and police officers would comply with such an order is exactly the point.

Richard Hanania's avatar

Soldiers and police officers are part of the government. So you're agreeing the government doesn't slaughter people for moral reasons.

Dean Valentine's avatar

My point is that right now, a large number of people (at least on a group level) have to consent to such atrocities, where as a much smaller number of people will be sufficient in an environment where power no longer rests in commanding the loyalty of actual human beings with guns.

Dave92f1's avatar

It doesn't matter if the gains are widely shared or not. Suppose the Tech Bros capture 99.9% of the gains, and toss the other 0.1% back as charity to everyone else (as unlikely as that scenario may be).

If that 0.1% is still enough to fund everyone else living in 100 acre palaces with robot servants and all having their private spaceships .... why do you care? Let the Tech Bros enjoy their planet-sized pleasure palaces. Don't be greedy or envious.

I think people who worry about the problem you mention don't understand what a singularity really is, or the scale of it. Don't use industrial age scarcity thinking where it doesn't apply.

M.J. Jacobi's avatar

The post-singularity techbros would still have a massive amount of power over everyone else in that example, and could easily wipe them out or drive them to living in the cracks and shadows (which is kinda what they'd be doing by default in such a lopsided situation). Though if I were a techbro I'd worry about superintelligence outcompeting and obsoleting me as well. Take any singularity scenario far enough and it starts becoming hard to see how human-like entities have any place in it.

Richard's avatar

It's more the socio-political ramifications that are a danger in that instance. In any case, like you, I doubt a cabal would own 99.9% of wealth, but (and this is one reason why authoritarian dictatorships are bad) you don't want a small cabal to ever capture 100% of all political decisions/decisions on violence.

Person Online's avatar

>It’s unlikely that the entire labor force ends up being basketball players. Some will play chess or Dungeons and Dragons. Others will do hobby-like or hobby-adjacent work like podcasting, playing video games, or streaming.<

Podcasting and streaming video games for a living are already industries unto themselves. We can't run an entire economy solely based on such activities because they require exponentially more consumers than producers in order to remain sustainable. A guy can't pull several thousand live viewers per day if every single potential viewer is already busy trying to run their own livestream.

The concern about AI and "jobs" is redistributive at the end of the day. No one with any sense is worried that the economy won't be productive--they are worried about having access to their slice of the pie, and rightfully so. This essay correctly points out that current redistributive policy makes no sense and redistributes wealth from those who, according to sense, generally need it the most (the young) to those who often need it the least (elderly retirees).

Why, then, would we expect redistribution in the face of an AI "jobspocalypse" to be any more rational or sensible? I imagine everyone would be pretty calm if they felt sure that AI and the end of "jobs" would actually result in the option to literally sit around and do nothing all day without working, but I see no reason to believe this will be the case, at least in the United States. There is not actually any US welfare program equating to this sort of "universal basic income" of which I am aware.

The typical American 18 year old is already facing a choice between living in mom's basement forever (assuming his parents are generous enough to give him this choice) or working part-time hours at a dead-end service job that he hates in order to earn enough money to barely make rent even after it is split between multiple roommates. His third option is to spend years and possibly many thousands of dollars in debt going through some kind of credentialing program, and it is this third option which is clearly and specifically threatened by AI. And of course the unspoken fourth option is simply homelessness.

Without an overhaul of redistribution policies, there is no clear line from this scenario to a scenario of infinite leisure where AI somehow allows this person sit around all day and do nothing and yet still afford rent, groceries, etc. The fact that a small minority of people are able to escape by becoming Internet influencers does not change the reality that the other ~95% of ordinary people must face.

Richard's avatar

No UBI right now, but we aren't talking about our current world. We're talking about essentially a post-materialistic one if AI really can do all current jobs. Americans living in the late 19th century couldn't have imagined a world where old people simply receive money for existing or free/subsidized healthcare for poor people or or even unemployment payouts from the government for being unemployed. Yet all those came to pass as the amount redistributed increased as overall wealth increased. Are you positing that redistribution wouldn't increase as aggregate wealth increases by 10X? 100X?

Person Online's avatar

It might, but all kinds of things might happen, emphasis on might. I don't feel particularly confident sitting here and just trusting that it will, and I think you'd be pretty stupid to bet your life plans on it, i.e. deciding to give up on college or quitting your job to live on savings because we'll all live in the post-materialist utopia soon anyways.

Richard's avatar

Yes, and Hanania didn't say a post-material world will definitely come about. Did you even read the article? He saying that -in the case- where AI is super-intelligent, and can do everything, we would certainly be in a post-materialistic world and there are far more other realistic risks to worry about than that the vast majority of people will live in destitute poverty. And if you have read any economic history, you'd agree. No country that has increased from late 19th century levels of aggregate wealth to current Western levels have kept redistribution at late 19th century levels (or close to it), so why would you think it's anywhere close to likely that a similarly large leap in aggregate wealth would lead to more destitution (same/less redistribution)?

The main thing I take away from this exchange is that you are like many people who are extremely lacking in imagination and thus can not comprehend a world that is much richer than the one we live in now (just like the typical person in the late 19th century in the US wouldn't have been able to comprehend how wealthy and well off the median American is now).

Kryptogal (Kate, if you like)'s avatar

The redistributive systems we have NOW only occurred bc 80 years ago the world had undergone a massive depression followed by an economy destroying global war and Americans with capital were terrified of the communist revolutions that had occurred and were this willing to give a but to avoid guillotines. Since the fall of the Soviet Union the taxation for decades has only gone in one direction: less taxes and less redistribution. I see no reason that anything would change or that they would be willing to redistribute a single dime absent threat of death. They've not shown themselves to be in the past, absent no other options.

Person Online's avatar

This is correct. Social upheaval in some form is a prerequisite to redistribution of wealth, and yet Richard would have us believe that everyone who gets up in arms in such a manner are crazy because somehow the wealth will magically flow down to them anyways.

Jason S.'s avatar

“The fact that humans will be materially fine in a world of aligned superintelligence does not necessarily mean that they will all be happy or fulfilled. We see famous actors and musicians die of drug overdoses and suicides at disturbingly high rates.”

This should one area of medicine among many where research, treatment and care can be supercharged in an exceptional productivity scenario. There are thousands of diseases detracting and even ruining people’s lives that can be targeted. How can this not be a high growth sector for employment?

Citizen Penrose's avatar

I think this argument requires inequality to be very high.

The problem with jobs that provide other people with niche luxuries, is that they can only be a large fraction of the total workforce if inequality is very high.

If you provide one other person with a luxury service, your income can only be the same as theirs if they pay a full half of their income for just that service. It's very unlikely anyone would want sacrifice half of their total consumption for just one additional item of consumption.

If a job provides many customers with a luxury service, there necessarily can't be many of that job, because each one requires a large customer base.

Theoretically you could have people of equal wealth trading luxury services between each other, like one trillionaire waits tables for another trillionaire on Mondays, who does the waiting on Tuesdays in return. But the time opportunity cost for trillionaires will be so high, no one would want that.

You can trade of income income inequality for employment if you want, but it creates other problems, like the distribution of positional goods (which will be really important post-scarcity).

Richard's avatar

In a post materialistic world, what would "opportunity costs" even be? The key rub is that not all people have the same interests/aptitude developing the same skills/knowledge/aptitude. So the yoga instructor who is really in to yoga would pay the travel consultant/guide who is really in to travel/exploring the world who would pay the dog whisperer who is really in to training dogs for oldog training who would pay the tutor who is really in to tutoring kids who would pay the yoga instructor for yoga classes.

Kryptogal (Kate, if you like)'s avatar

No one pays travel consultants now!! And the only people who pay for any of these things you mentioned are already in the UMC and making far, far more money than 80% of people. Like I pay a dog hiker and for an occasional massage and now and then for food delivery and some Substackers, but my income is in the top 5% and I fired my house cleaner three years ago bc it cost too much and if my income went down even 15% I would stop paying for all of those things.

Richard's avatar

If AI can do any job, it’s pretty much inconceivable that in such a world, there would not be a massive amount of wealth created and that the vast majority of people would not be as rich as you are now.

Kryptogal (Kate, if you like)'s avatar

I've never known rich people to voluntary share or redistribute absent no other option, and sometimes they don't even do it when they're being dragged to the guillotine, so I disagree. I think the people with assets will be perfectly happy to live in luxury compounds while everyone else lives like a feral animal. See eg most shanty towns right outside massive wealth in many cities across the globe.

Argentus's avatar

There are two tendencies getting in the way of being practical on this topic. One is "cut down the tall poppies" status anxiety. The other is paranoia about freeloaders.

To actually have anything like this future you must simultaneously accept:

1. It's okay if uber elites have their private luxury planets.

2. It's also okay if stupid, ugly, lazy, immoral slobs have high-quality lives *including* high-quality leisure and enjoyment. (They too can have champagne and caviar and not just nutrient paste to stop them from starving).

Put another away - you must get over your lizard brain disgust/outrage/fear at the mere existence of the bourgeoise and/or the lumpenproletariat.

This is very hard because 1) there are evolutionary underpinnings for both of these impulses and 2) one type of disgust is load-bearing on the right and the other is load-bearing on the left.

M.J. Jacobi's avatar

Well, the thing about #1 is it implies the elites would have a lot of power over regular people too (the price of a planet ought to buy a very large automated army). But maybe we'll all be equally powerless before the AI gods.

DeepLeftAnalysis🔸's avatar

"The idea that all mechanisms of redistribution completely disappear in a world with superintelligence is part of the denial of human nature that is required for one to fear the jobspocalypse scenario."

I'm pro-AI, but the anti-AI argument is that the economy will no longer be ruled by human nature.

Richard's avatar

How could that be true when humans still control power? Do you think human politicians will let AI make the ultimate decision on what policies to support? That human rich people will let AI make the ultimate decision on where to give money? That human voters will let AI decide for them who to vote for?

M.J. Jacobi's avatar

Humans will cede power to the AIs the moment it becomes possible, look at how many people are outsourcing their thinking to it already. Those who don't will be outcompeted. If the Singularity happens, all I can really hope is the Ascended Economy still has room for humans in some nook or cranny.

DeepLeftAnalysis🔸's avatar

That's the argument that needs to be addressed, yes.

El Toro's avatar

A major hole in the optimist narrative is that there is no credible evidence that the low hanging fruit of ASI would be "post-scarcity" or "hyperabundance", which are sci-fi tropes that may or may not be possible in reality. There is evidence that ASI may offer many advantages in unregulated zero sum competitions, cyberwarfare being the most obvious example, but other types of warfare including mass destruction could soon follow. If this involves the deployment of poorly aligned ASI by the increasingly unwise and unaccountable world elites, things might get ugly.

Nishant Juneja's avatar

I think he pretty explicitly is assuming "alignment" (which, certainly, is a bit of a handwavey assumption) in this optimistic take.

El Toro's avatar

He's making two major assumptions. One is a very strong form of alignment. I personally believe that alignment is plausible if ASI developers prioritize it, but at the moment it's an afterthought to them. The other wildly speculative and unsupported assumption is that an aligned ASI leads to hyperabundance. Energy and natural resources could become a critical bottleneck.

Richard's avatar

He's not assuming abundance. He's arguing that super-AI is equivalent to abundance. As for strong alignment, read the article! Hanania explicitly said he thinks us all being killed/enslaved by AI is a bigger risk than that we have a teeming mass of destitute jobless in a world of super-AI.

But in any case, when it comes to alignment, human beings have a STRONG incentive to make sure that AI (super or otherwise) is aligned, and incentives are almost never beaten.

The EVIL Lich's avatar

There is only a limited number of resources, and only so far that they can be stretched with labor. What makes you think AI and corporations holding ownership over AI systems won't be able to outcompete humans for those resources? How will I buy food if all the farmers are using their fields to grow bio-fuel for artificial intelligence systems, which can offer much more value in exchange than I can?

Humans may have unlimited wants, but have limited time. Why would someone watch Latina grandmas play basketball, when they can instead watch an AI generated video, algorithmically guaranteed to produce more entertainment?

ApartofTheHumanRace's avatar

The thing is: back then, whenever an automation shock happened there were a variety of jobs from a higher pay sector for a technologically displaced person to persue (Farmhands can get higher-paying jobs in manufacturing sector, Factory workers can get higher-paying service sector jobs) but with AI (at the very least LLMs) the shock is going to be felt by well off white collar workers and there really isn't a path available for them for upward mobility. A coder who's rendered redundant would have to get a new job which not only pays lower but a job which they're most likely very overqualified for since it's a specialised area if expertise. The same goes for nurses and accountants.

Maybe there are deficits in my mental model (ie they could get jobs in the skilled trade and crafts sector which might have similar pay, but it also assumes vocational training is quick and easy, which is not) but I feel like for the current context, AI might cause a larger shock in employment than what came before. Plus, with the current media landscape, any indignities felt towards automation will be amplified and we might have a luddite movement larger than the original one which happened in the 19th century.

Worley's avatar

These sorts of arguments are quite true ... Consider what happened when a lot of well-paying blue collar jobs were eliminated. Many such people moved into the hospitality industry. When I have dinner at the neighborhood bar and grill, as much as 1/3 of what I pay is actually to have the waiter tend to me. From a purely rational point of view, that is just wasted action, but still I pay for it. Similarly, it's possible to imagine an "artisanal" economy where all conceivable material needs are available at infinitesimal costs and everybody spends their money paying for humans to do for them things that machines would do for nearly free ... and they get that money by doing artisanal things for other humans.

But the negative is that *which* skills pay well may change drastically. With the invention of spinning mills, the hand weavers were replaced with the loom attendants, which needed somewhat different skills. The invention of the backhoe replaced the ditch-digger and physical strength was no longer particularly valuable. Over the past few decades, high cognitive capabilities were very valuable and such people (like you and I and all of Hanania's readers) moved up the status ladder. AI is likely to render pure cognitive capability much less valuable whereas e.g. the ability to be a pleasing waiter may well move up the status ladder.

James Alexander's avatar

For Latina grandmother basketball stars read dog and cat psychotherapists...

Sohale Mortazavi's avatar

Isn’t so much wealth transferred to olds because they secured those programs for themselves when labor had far more leverage than workers do now?

Worley's avatar

More like they've secured it because they vote at higher rates ...

Sohale Mortazavi's avatar

That’s definitely (somewhat) helpful in protecting programs from cuts, but Medicare and Social Security weren’t implemented by seniors for seniors. They were created by society at large at a time when labor had more political power and leverage.

Shane Watkins's avatar

I was told before that the Luddites were correct in that their actual living standards decreased. Transition costs and what not made them worse off, while enriching society generally.

Pessimism about the AI revolution taking jobs is from the people that will be displaced. Journalists and editors will lose status as the machines will make the skills and connections they've made obsolete. That's where all this pushback is coming from. These people don't want to lose status.

Worley's avatar

> Luddites were correct in that their actual living standards decreased.

That's true, but the median standard of living in England increased. The difference was that the class of hand-weavers moved down the class rankings severely (and by mathematical necessity, everybody else moved up a bit).

alesziegler's avatar

Hello, I am here to nitpick obnoxiously. Nobody would be killed if we banned psychiatry?? Really?? Or accounting?? Oldest accounting tables known are from ancient Sumer. Running a modern civilization without accounting is obviously impossible, and I can guarantee that without modern civ, life expectancy would be drastically shorter.

Of course you could make a list of non-critical professions, but these were a very poor choices.

Richard's avatar

Yeah, I understood Hanania's, but he strawmanned/exaggerated.