AIs gunning for our precious freelancers

[Ketsuban]

Sure but eventually everyone becomes unemployable and indeed superfluous to the running of civilization.

I want to point out the size of the gulf you've just vaulted, Evel Knievel-style. The quoted post is talking about the notoriously vibes-based (to use the modern vernacular) process in which a radiologist looks at a scan and makes an imperfect best guess based on having looked at lots of scans in the past as to whether or not it shows cancerous growth. The accuracy of this process is improved by the addition of computer vision as a second opinion despite it also not being 100% accurate. However, you've leapt from that to "but what if there are no more jobs at all". As such…

Nobody advocating AI has a good answer to this problem apart from pinky swearing to keep us around as pets once all jobs have been eliminated.

…what you're asking about is such an absurdly distant possible future scenario that they see no need to take it seriously, any more than someone inventing the safety match should need to address the question of what happens if their matches burn all the oxygen in the atmosphere.

[zompist]

Hypothetically, there's an interesting conworlding question -- if we decided to get rid of all post-1800 technology, are we sure we would get lords and peasants again? (To be clear: I don't want that! People would die!)

To the extent that you can describe any pre-1800 society as "lords and peasants", then yes. But there's a lot of complications.

Premodern societies were run in various ways. The Chinese elite was scholar-officials— though make no mistake, these were rich self-perpetuating elites. There were republics— always oligarchies— from Greece to early Rome to Italy to India to Kiev to Tlaxcala. James Scott would point out that as much of a third of humanity lived outside traditional states, either as nomads, hunter-gatherers, or in non-grain-producing areas that never really submitted to states.

Europe had a large middle class, several hundred years old by 1800. As Graeber & Wengrow point out, the French Academy— not exactly a bastion of revolutionary thought— posed the debate question "What is the origin of inequality among men, and is it authorized by natural law?” in 1754: Europeans were already debating equality, largely due to Native North American examples. There were major democratic revolutions before 1800.

Also, if you had the conditions of 1800, why would you not, in twenty years time, have the conditions of 1820? The industrial revolution was already underway; the Royal Society and the Académie des Sciences had been collating scientific work for 140 years; the steamboat, photography, and the railroad were invented in that 20-year period and were probably inevitable. Inventing a society that stays at 1800 level requires some sort of suppressive mechanism, and that sure isn't coming from the peasants.

Could you have some kind of hippie commune version of the 1800s without lords? Sure, see the Native North Americans again. But it's not accident that they were pushed out of their lands by Europeans— more by disease and demographics than by guns. The secret weapon of peasant states is the womb. (Though even sf/fantasy writers, to say nothing of emperors, forget that premodern agricultural states can't expand into grassland, desert, wetlands, or mountain zones.)

(I'm putting aside all questions of quality of life, which obviously would plummet. And I'm sure everyone here knows this, but in 1800 the villain of the day was not "technology" or "capitalism"; it was mercantilist imperialism. There were proto-capitalists, but they were not in power, and capital did not function as it does today until the invention of the telegraph. Adam Smith pooh-poohs the notion of the corporation.)

[malloc]

…what you're asking about is such an absurdly distant possible future scenario that they see no need to take it seriously, any more than someone inventing the safety match should need to address the question of what happens if their matches burn all the oxygen in the atmosphere.

Sure but things can change incredibly fast. Remember my example with airplanes advancing from crude prototypes to military weapons in only one decade. Obviously that hardly proves that AGI is just around the corner, but the "distant" future has a way of arriving when we least expect it. We need precautions for the future, not belated half-measures after the crisis has already become undeniable. For that matter, any AI smart enough to replace radiologists could also replace many other jobs of comparable intellectual difficulty. What seems like minor improvements in one field could easily translate to massive and sudden collapse in employment more generally.

[Travis B.]

…what you're asking about is such an absurdly distant possible future scenario that they see no need to take it seriously, any more than someone inventing the safety match should need to address the question of what happens if their matches burn all the oxygen in the atmosphere.

Sure but things can change incredibly fast. Remember my example with airplanes advancing from crude prototypes to military weapons in only one decade. Obviously that hardly proves that AGI is just around the corner, but the "distant" future has a way of arriving when we least expect it. We need precautions for the future, not belated half-measures after the crisis has already become undeniable. For that matter, any AI smart enough to replace radiologists could also replace many other jobs of comparable intellectual difficulty. What seems like minor improvements in one field could easily translate to massive and sudden collapse in employment more generally.

The future that happens is not always the future that we predict. I have on my table microcontroller boards that cost $7 in today's money that are much more powerful than multimillion USD (in 1960's money) mainframes from the 1960's that can communicate using the Internet with countless machines across the world if I so desire, something that people back in the 1960's could barely conceive of. Yet at the same time, we still haven't returned to the moon.

[keenir]

While they often raise legitimate points like the racism of Immanuel Kant or the hardships of capitalist work culture, it feels like they gloss over the sheer drudgery of subsistence farming or the intellectual limitations of pre-scientific thought.

but how much do the personal preferences of philosophers, actually mandate or shape the ways of life being suggested in these discussions? I mean, even the harshest critic of nuclear power, doesn't say "and look, Einstein married his cousin!"

Despite their cosseted existence, pets fundamentally have no control over their lives.

depends on the owner.

[keenir]

Sure but things can change incredibly fast. Remember my example with airplanes advancing from crude prototypes to military weapons in only one decade. Obviously that hardly proves that AGI is just around the corner, but the "distant" future has a way of arriving when we least expect it. We need precautions for the future, not belated half-measures after the crisis has already become undeniable. For that matter, any AI smart enough to replace radiologists could also replace many other jobs of comparable intellectual difficulty

No, an AI able to replace radiologists, can only replace radiologists. you keep underestimating how specialized AI are -- an AI that can draw a satisfactory picture, is not one that can weigh in on voting (ir)regularities or population health.

[keenir]

…what you're asking about is such an absurdly distant possible future scenario that they see no need to take it seriously, any more than someone inventing the safety match should need to address the question of what happens if their matches burn all the oxygen in the atmosphere.

Sure but things can change incredibly fast. Remember my example with airplanes advancing from crude prototypes to military weapons in only one decade.

and remember that airplanes had design improvements and plannings coming from two directions: the humans in the design barns, and the humans operating the planes. (and also humans elsewheres)

AI do not have all of those.

[malloc]

No, an AI able to replace radiologists, can only replace radiologists. you keep underestimating how specialized AI are -- an AI that can draw a satisfactory picture, is not one that can weigh in on voting (ir)regularities or population health.

Sure the specific AI would not necessarily take over other professions, but successfully automating one profession with AI provides a roadmap for repeating the process with others.

[keenir]

No, an AI able to replace radiologists, can only replace radiologists. you keep underestimating how specialized AI are -- an AI that can draw a satisfactory picture, is not one that can weigh in on voting (ir)regularities or population health.

Sure the specific AI would not necessarily take over other professions, but successfully automating one profession with AI provides a roadmap for repeating the process with others.

suuuure, just like creating Klingon is just like creating Toki Pona. Inventing a radio controlled toy car is exactly like making a nuclear sub.

also, if you have an AI performing surgeries, you haven't automated surgery -- you have replaced one intelligence with a supposed intelligence. the way I learned the term, thats not automating the process...no more than a surgeon using waldoes, would be automation.

[Ares Land]

French company MistralAI https://mistral.ai/en/news/le-chat-mistral reached a breakthrough as well; their model is apparently as efficient as DeepSeek, and is also open source.

I'm still not sold on the interest of generative AIs, but at least this one is pretty modest in terms of energy requirements, also it's apparently hosted in France (which means the carbon footprint is very low, even if, alas, the energy mix is mostly nuclear here ) Interesting that this was made for a fraction of the Project Stargate budget.

[alice]

@malloc: what would it take to prove you wrong, or to convince you that your worries have been unfounded?

[malloc]

@malloc: what would it take to prove you wrong, or to convince you that your worries have been unfounded?

I would need good reason to believe that AI development will soon stagnate before it encroaches on more human activities. Currently it seems like one job after another has AI projects aimed at eliminating it and many are working on artificial general intelligence that can replace humans in all areas.

also, if you have an AI performing surgeries, you haven't automated surgery -- you have replaced one intelligence with a supposed intelligence. the way I learned the term, thats not automating the process...no more than a surgeon using waldoes, would be automation.

Practically speaking you have automated the process because that intelligence doesn't require payment or time off, which is what the people replacing workers with AI really want.

[keenir]

@malloc: what would it take to prove you wrong, or to convince you that your worries have been unfounded?

I would need good reason to believe that AI development will soon stagnate before it encroaches on more human activities

except every time we provide such good reasons, you dismiss them.

also, if you have an AI performing surgeries, you haven't automated surgery -- you have replaced one intelligence with a supposed intelligence. the way I learned the term, thats not automating the process...no more than a surgeon using waldoes, would be automation.

Practically speaking you have automated the process because that intelligence doesn't require payment or time off, which is what the people replacing workers with AI really want.

except that what they want =/= what automation is.

[malloc]

except every time we provide such good reasons, you dismiss them.

You have suggested challenges with existing models perhaps, but no reason why AI companies couldn't solve those limitations or even develop whole new models without them.

except that what they want =/= what automation is.

If the AI has achieved sentience and thus subjecthood, then I suppose that would make sense. Instead of eliminating a worker, you have simply replaced one worker with another of a different species (or whatever you would call artificial intelligences). However most people have assured me that AI has a long road to sentience and it seems unlikely that a surgical robot would require that level of functionality.

[keenir]

except every time we provide such good reasons, you dismiss them.

You have suggested challenges with existing models perhaps, but no reason why AI companies couldn't solve those limitations or even develop whole new models without them.

thats like arguing that commercial airlines don't need all those long runways -- after all, there is no reason why plane companies can't solve those limitations or develop new planes that don't need them.

except that what they want =/= what automation is.

If the AI has achieved sentience and thus subjecthood, then I suppose that would make sense. Instead of eliminating a worker, you have simply replaced one worker with another of a different species (or whatever you would call artificial intelligences). However most people have assured me that AI has a long road to sentience and it seems unlikely that a surgical robot would require that level of functionality.

who are these "most people" that you listen to the opinions of? every time anyone here tries to tell you something, you ignore or dismiss them.

[sasasha]

I'm going to throw a bomb in here. Sorry.

I am with malloc – in that I think AI will eventually bring about profound changes to human society, and eventually rework the structures of how stuff gets done, how people spend their time, and how they participate in resource accumulation/administration (for want of a word that is not 'the economy' because I don't think it will be recognisable as such).

However, I am emphatically not with malloc – in that I don't think it's anything we can, or should try to fight against.

Malloc's objections seem to me like the priesthood of a pre-literate society shaking their heads at this awful new writing stuff. Yes, orality (in terms of knowledge-keeping and propogation) has kinda gone down the pan since we started writing things down. Yes, people can no longer make a living specifically being rememberers. But no, I don't think humanity is worse off as a result. Literacy opened an enormous trove of human endeavour and activity and so will AI.

Yes ethical issues do and will abound. Such is life. Yes AI technology will progress in unpredictable ways. Such is the inevitable dance of life.

I live in a former textiles town. One of the first places in the world to have huge mills with thousands of employees. Children as young as 8 were employed to work 12 hour days in frighteningly dangerous conditions. It wasn't pretty whilst the ethical lines were still being worked out but now we have... well... stuff billions of people can wear, and respectable textiles professions, and employment law, and work-life balances. (And people on the other side of the planet still working in shitty conditions for our market, unfortunately – but maybe AI-run stitching machines will one day free them up too?).

I'm sorry but these sorts of teething problems sort themselves out in the end. I think AI is a step in the road to a post-scarcity society. It will be a bumpy stairway. But I prefer post-scarcity to scarcity, honestly. Tell anyone who's starving different.

Sorry if this seems hopelessly utopian, or the opposite. In summary, I am not blind to the issues malloc is raising, but such issues are just part of the human condition. Responsible engagement with them is not, in my opinion, shouting to the sky and wishing that it were not so. It is so, AI is an enormously powerful tool and it is now here. We need to learn to live with, not against as well and as ethically and as optimistically as we can.

[bradrn]

Related to sasasha’s comments, I saw this random post on Hacker News, which really resonated with me:

Was the invention of the printing press a net good for humanity? Most certainly so, looking back from today. Did people living back then knew what they were getting into? Not really. And since their share of the fruits of that invention was mostly bloodshed, job loss, and shattering of the world order they knew, I wouldn't blame them from being pissed off about getting the short end of the stick, and perhaps looking for ways to undo it.

I'm starting to think that talking about inventions as good or bad (or the cop-out, "dual use") is bad framing. Rather, it seems to me that every major invention will eventually turn out beneficial, but introducing an invention always first extracts a cost in blood. Be it fire or printing press or atomic bomb, a lot of people end up suffering and dying before societies eventually figure out how to handle the new thing and do some good with it.

I'm very much in favor of progress, but I understand the fear. No matter the ultimate benefits, we are the generation that cough up blood as payment for AI/AGI, and it ain't gonna be pleasant.

Some of the specific claims here (especially those which I didn’t quote) are perhaps dubious, but the overall point seems basically correct to me.

[sasasha]

Related to sasasha’s comments, I saw this random post on Hacker News, which really resonated with me:

Was the invention of the printing press a net good for humanity? Most certainly so, looking back from today. Did people living back then knew what they were getting into? Not really. And since their share of the fruits of that invention was mostly bloodshed, job loss, and shattering of the world order they knew, I wouldn't blame them from being pissed off about getting the short end of the stick, and perhaps looking for ways to undo it.

I'm starting to think that talking about inventions as good or bad (or the cop-out, "dual use") is bad framing. Rather, it seems to me that every major invention will eventually turn out beneficial, but introducing an invention always first extracts a cost in blood. Be it fire or printing press or atomic bomb, a lot of people end up suffering and dying before societies eventually figure out how to handle the new thing and do some good with it.

I'm very much in favor of progress, but I understand the fear. No matter the ultimate benefits, we are the generation that cough up blood as payment for AI/AGI, and it ain't gonna be pleasant.

Some of the specific claims here (especially those which I didn’t quote) are perhaps dubious, but the overall point seems basically correct to me.

Yes. It's a whimsical but generally effective way to put it. I was thinking of using "Sometimes it falls to a generation to be great" somewhere in this discussion. I'm not really into blindly aggrandising the process of assimilating a powerful new technology into society, as that seems rather callous given the suffering that will be involved, but I'm trying to find an expression of what I see as our lot in life to do so that sits somewhere, tonally, between that quote and "we are the generation that cough up blood as payment".

[bradrn]

Yes. It's a whimsical but generally effective way to put it. I was thinking of using "Sometimes it falls to a generation to be great" somewhere in this discussion. I'm not really into blindly aggrandising the process of assimilating a powerful new technology into society, as that seems rather callous given the suffering that will be involved, but I'm trying to find an expression of what I see as our lot in life to do so that sits somewhere, tonally, between that quote and "we are the generation that cough up blood as payment".

Agreed, yes. Perhaps something like: 'it is our generation's responsibility'.

[malloc]

Malloc's objections seem to me like the priesthood of a pre-literate society shaking their heads at this awful new writing stuff. Yes, orality (in terms of knowledge-keeping and propogation) has kinda gone down the pan since we started writing things down. Yes, people can no longer make a living specifically being rememberers. But no, I don't think humanity is worse off as a result. Literacy opened an enormous trove of human endeavour and activity and so will AI.

You can hardly compare literacy to AI for the simple reason that letters have no autonomy or opacity. The inventor of a writing system decides what each letter means and subsequent writers choose the letters they want on the page to convey their intended meaning. Artificial intelligence by contrast makes decisions we cannot predict or even understand. It takes over the processes of thinking and planning from humans and reduces us to mere supplicators with no real control over the process.

I'm sorry but these sorts of teething problems sort themselves out in the end. I think AI is a step in the road to a post-scarcity society. It will be a bumpy stairway. But I prefer post-scarcity to scarcity, honestly. Tell anyone who's starving different.

Meanwhile vast swathes of the population will find themselves permanently unemployed because AI has taken over so much of the economy. Anyone who thinks the oligarchs who currently run the world will enact UBI to support all these people is delusional to the point of psychosis. For that matter, the people designing and promoting AI are themselves part of that oligarchy and AI is clearly part of their plan to immiserate the rest of us rather than lifting us out of poverty.

Sorry if this seems hopelessly utopian, or the opposite. In summary, I am not blind to the issues malloc is raising, but such issues are just part of the human condition. Responsible engagement with them is not, in my opinion, shouting to the sky and wishing that it were not so. It is so, AI is an enormously powerful tool and it is now here. We need to learn to live with, not against as well and as ethically and as optimistically as we can.

We also have the option of deciding it's not worth the risk and quashing it decisively. Eugenics once enjoyed the same level of messianic enthusiasm that now characterizes artificial intelligence. After the horrors of Nazism however, people concluded that it was simply too dangerous and it ultimately became taboo.