Zompist Bboard Again

alice wrote: ↑Sun Feb 18, 2024 2:11 pm Here's something to make you reconsider; AI-generated images of cells:

https://freethoughtblogs.com/pharyngula ... f-inspire/

A few more at the reddit link. Perhaps some inspiration for very low-level biological conworlding???

Raphael wrote: ↑Wed Feb 21, 2024 2:53 am I try to avoid falling into the classical "the technology invented between your birth and your thirtieth birthday is amazing; the technology invented afterwards is a sign of everything wrong with the world"-trap, but, well, I do have the impression that almost all the really impressive and useful technology invented during my lifetime so far was invented before I turned 30 - that is, before 2012 - and that most or all of the stuff invented since then is either completely useless or more harmful than useful.

I think the image generation technology itself is beneficial. What is harmful is the market mechanism that can be misused to put human artists out of business. As alynnidalar mentioned, this technology allows writers to get illustrations made cheaply. DALL-E came up with all of these from fairly generic prompts: https://drive.google.com/drive/folders/ ... sp=sharing

alynnidalar wrote: ↑Fri Sep 02, 2022 11:56 am Directly with regards to AI creating art... right now, as a person of limited artistic ability and limited access to art supplies, the only way for me to get a custom piece of artwork--with obvious benefits of improving the conditions I live in, holding important personal meaning for myself, etc.--is to pay a large amount of money to have it commissioned. Much like your fan wavers, only the upper echelons realistically can do this.

With AI-generated art, custom art and design becomes accessible to anyone with access to the internet--not even a computer, a smartphone will do just fine! No longer can only the wealthy have access to beautiful custom artwork that suits their tastes and needs, but it is now widely available to the general public. Furthermore, there is an enormous functional difference between the lengthy process of communicating with an artist, refining your vision, etc. compared with the quick response time and direct tuning I can do with an AI artwork generator.

(you might say that having custom art is a bit less pressing of a human need than being able to function in hot weather, and yes, obviously I'm exaggerating and engaging in a dose of devil's advocacy to make my point, but on the other hand if you criticize me for that, I'll retort that the invention of air conditioning was clearly a capitalist plot to extract more work out of ordinary members of society during summer months, and should not our goal be to enable people to enjoy artistic pursuits rather than drudgery for The Man, so what do you think of that)

(seriously though--I don't think there has to be a specific "need" before people are allowed to experiment with things. Many many many useful (and artistic!) things are the result of people playing around, having happy accidents, not specifically pursuing the thing that they ended up with. Penicillin, teflon, Scotchgard, etc..... I don't like this idea that things must be Useful TM before we are allowed to be interested in them)

rotting bones wrote: ↑Wed Feb 21, 2024 9:45 pmI think the image generation technology itself is beneficial. What is harmful is the market mechanism that can be misused to put human artists out of business. As alynnidalar mentioned, this technology allows writers to get illustrations made cheaply. DALL-E came up with all of these from fairly generic prompts: https://drive.google.com/drive/folders/ ... sp=sharing

Beneficial to whom? You can say that businesses should only use the technology for non-artistic purposes like illustrating research papers, rather than replacing human artists in entertainment. But actually enforcing that policy sounds rather difficult in practice, requiring considerable levels of red tape to restrain wealthy businesses with armies of lobbyists and the average person alike. Imagine regulators watching your internet usage to ensure that you get your dragon-girl OC drawn by a human artist rather than prompting midjourney.

The fundamental problem is that humans have numerous staggering disadvantages compared with artificial intelligence. Homo sapiens is merely an ape, albeit highly derived. We have all the weaknesses and grotesqueries of other apes: lust and tribalism, all manner of irrationality, legions of bacteria in every orifice and even inside our guts, all while lacking their physical strength and hardiness. Artificial intelligence by contrast is pure intellect unsullied by fleshly desires or irrationality. Computers don't fart or check on their spouses nor do they trade the inviolable laws of arithmetic for "pi equals four" because their religion demands it. Employers now have the choice of hiring either orcs or elves and unless they're Sauron, the former has little to offer.

[One thing that frustrates me about this debate is that numerous people assert that AI cannot achieve sentience or general intelligence without explaining why. Cory Doctorow compares the development of AGI to selectively breeding racehorses into locomotives. Yet he fails to consider that the data centers hosting AI already are analogous to locomotives: massive machines with incredible power behind them. Conversely, humans are more analogous to horses: organic beings who evolved to thrive on grasslands yet have incredibly flimsy constitutions. Furthermore, the only reason anyone would want to breed horses into trains is improving their ability to transport goods and trains already beat horses in that area by many orders of magnitude.]

malloc wrote: ↑Thu Feb 22, 2024 8:43 am Beneficial to whom? You can say that businesses should only use the technology for non-artistic purposes like illustrating research papers, rather than replacing human artists in entertainment. But actually enforcing that policy sounds rather difficult in practice, requiring considerable levels of red tape to restrain wealthy businesses with armies of lobbyists and the average person alike. Imagine regulators watching your internet usage to ensure that you get your dragon-girl OC drawn by a human artist rather than prompting midjourney.

Image generation is beneficial to poor writers without the means to hire illustrators. Large corporations are the ones who can afford to commission bespoke art by hiring human illustrators. (Traditionally, "liberals" come in at this juncture and call small creators losers opposing social justice, throwing the election to convicts and Nazis.)

You cannot enforce social justice by stifling science. My solution is to boldly make the private sector poorer and less efficient by providing art studios with government support through a popular vote. Politicians are structurally incapable of bringing this about because of the Prisoner's Dilemma that's driving their race to the bottom. Therefore, we should abolish politicians as a profession and have direct democracy instead.

malloc wrote: ↑Thu Feb 22, 2024 8:43 am One thing that frustrates me about this debate is that numerous people assert that AI cannot achieve sentience or general intelligence without explaining why.

They can. The problem is that a sentient being is not a more powerful calculator along any straightforward line of development. Sentience is a different functionality, one that is unprofitable for businesses to put in charge of their work. Their computers joining a strike would hurt the corporation's bottom line.

rotting bones wrote: ↑Thu Feb 22, 2024 1:20 pmImage generation is beneficial to poor writers without the means to hire illustrators. Large corporations are the ones who can afford to commission bespoke art by hiring human illustrators.

Except that large corporations no longer need to hire human artists because computers can draw now. It hardly makes sense for them to continue employing humans for something they can have for free. Between penny pinching corporations and people too poor to hire artists, humans no longer have any viable career in drawing. Certainly it sucks that many people cannot afford the drawing of their dreams but replacing human artists with generative AI seems a steep price to solve that. It sounds like you genuinely don't put much stock in the humanities and wouldn't lose much sleep if they disappeared entirely. But realistically, where does that leave humans in terms of finding employment or simply participating in civilization as anything other than mindless consumers?

They can. The problem is that a sentient being is not a more powerful calculator along any straightforward line of development. Sentience is a different functionality, one that is unprofitable for businesses to put in charge of their work. Their computers joining a strike would hurt the corporation's bottom line.

Sure but plenty of people seem adamant that computers cannot achieve sentience nor even general intelligence. Indeed it seems like pretty much every critic of the tech industry I've encountered (Paris Marx, Cory Doctorow, and so forth) agrees on this point. Doctorow repeatedly derides the notion of developing AGI as breeding racehorses into locomotives in his blogs.

malloc wrote: ↑Thu Feb 22, 2024 2:31 pm Except that large corporations no longer need to hire human artists because computers can draw now. It hardly makes sense for them to continue employing humans for something they can have for free. Between penny pinching corporations and people too poor to hire artists, humans no longer have any viable career in drawing.

That's not what will happen. Under capitalism, rich consumers will continue to pay for luxury products with bespoke art because they can afford it. These luxury products can be produced by corporations, writers rich enough to hire illustrators, small "artisanal" companies, etc. Most consumers will make do with AI art most of the time.

malloc wrote: ↑Thu Feb 22, 2024 2:31 pm Certainly it sucks that many people cannot afford the drawing of their dreams but replacing human artists with generative AI seems a steep price to solve that. It sounds like you genuinely don't put much stock in the humanities and wouldn't lose much sleep if they disappeared entirely. But realistically, where does that leave humans in terms of finding employment or simply participating in civilization as anything other than mindless consumers?

I'm not sure how you could have missed that I'm proposing a drastic restructuring of society for the sake of art:

My solution is to boldly make the private sector poorer and less efficient by providing art studios with government support through a popular vote. Politicians are structurally incapable of bringing this about because of the Prisoner's Dilemma that's driving their race to the bottom. Therefore, we should abolish politicians as a profession and have direct democracy instead.

These art studios would employ high quality human artists. They would receive popular support because there is demand for their products among people who don't have the money to pay for them in a capitalist market. Instead of a profit-driven money system, demand would be measured by equal votes.

malloc wrote: ↑Thu Feb 22, 2024 2:31 pm Sure but plenty of people seem adamant that computers cannot achieve sentience nor even general intelligence. Indeed it seems like pretty much every critic of the tech industry I've encountered (Paris Marx, Cory Doctorow, and so forth) agrees on this point. Doctorow repeatedly derides the notion of developing AGI as breeding racehorses into locomotives in his blogs.

I don't read them. I read the research on Google Scholar and aggregator sites like MarkTechPost: https://www.marktechpost.com

rotting bones wrote: ↑Thu Feb 22, 2024 2:40 pmThat's not what will happen. Under capitalism, rich consumers will continue to pay for luxury products with bespoke art because they can afford it. These luxury products can be produced by corporations, writers rich enough to hire illustrators, small "artisanal" companies, etc. Most consumers will make do with AI art most of the time.

So only most artists will lose their jobs, not all, and the few remaining ones will only survive as conspicuous consumption for the rich. That still sounds pretty horrible to me, even if not quite the artistic apocalypse. Certainly it does little to endear me to generative AI since the difference between 1% and 0% of artists remaining still sounds pretty small. The question is not whether some handful of obsequious artists will find wealthy patrons but why the rest should have to lose their livelihood and source of joy just so wealthy techies can get even richer.

I'm not sure how you could have missed that I'm proposing a drastic restructuring of society for the sake of art:

My solution is to boldly make the private sector poorer and less efficient by providing art studios with government support through a popular vote. Politicians are structurally incapable of bringing this about because of the Prisoner's Dilemma that's driving their race to the bottom. Therefore, we should abolish politicians as a profession and have direct democracy instead.

Not missed so much as ignored since I am focusing on what seems likely for the future rather than what ought to happen. One can easily imagine scenarios where radical political changes have lessened the harm of automation, but I still have to worry about what's coming in the real world.

malloc wrote: ↑Thu Feb 22, 2024 4:13 pm So only most artists will lose their jobs, not all, and the few remaining ones will only survive as conspicuous consumption for the rich. That still sounds pretty horrible to me, even if not quite the artistic apocalypse. Certainly it does little to endear me to generative AI since the difference between 1% and 0% of artists remaining still sounds pretty small. The question is not whether some handful of obsequious artists will find wealthy patrons but why the rest should have to lose their livelihood and source of joy just so wealthy techies can get even richer.

Is it that dire? The market for luxury products looks huge to me. Lots of people use luxury products for certain types of articles and bargain bin items for everything else. That's largely what defines one's personality (AKA social media profile) under consumerism.

malloc wrote: ↑Thu Feb 22, 2024 4:13 pm Not missed so much as ignored since I am focusing on what seems likely for the future rather than what ought to happen. One can easily imagine scenarios where radical political changes have lessened the harm of automation, but I still have to worry about what's coming in the real world.

If radical political change is impossible, then what are you even discussing? Whatever will happen will just happen, and that's what will happen. The system cannot be changed, generative AI research will happen, and jobs will be lost. Period.

For me to side against AI research, I would have to value human art more than AI research, and I don't. Nevertheless, I do value human art, and my compromise is to support government jobs for human artists.

rotting bones wrote: ↑Thu Feb 22, 2024 5:07 pmIf radical political change is impossible, then what are you even discussing? Whatever will happen will just happen, and that's what will happen. The system cannot be changed, generative AI research will happen, and jobs will be lost. Period.

Mostly, I am lamenting the horrors to come and hoping for others to acknowledge them instead of telling me that art was never that important anyway. Also, reforms and regulations are possible even if revolution seems hopelessly remote at this point. Until it becomes feasible to overthrow capitalism, though, we need to safeguard humanity against threats like global warming or AI destroying our livelihoods or worse.

For me to side against AI research, I would have to value human art more than AI research, and I don't. Nevertheless, I do value human art, and my compromise is to support government jobs for human artists.

But why? How can you value the interests of future machines over the livelihood and dignity of humans living now? Technology ought to serve the interests of humanity and yet so many tech workers are hellbent on sacrificing humanity for the sake of AI. Are you embarrassed by your simian heritage and wish to see genus Homo extinguished once and for all, or do you simply find AI so fascinating as a concept that it supersedes human interests?

malloc wrote: ↑Thu Feb 22, 2024 5:41 pm But why? How can you value the interests of future machines over the livelihood and dignity of humans living now? Technology ought to serve the interests of humanity and yet so many tech workers are hellbent on sacrificing humanity for the sake of AI. Are you embarrassed by your simian heritage and wish to see genus Homo extinguished once and for all, or do you simply find AI so fascinating as a concept that it supersedes human interests?

AI researchers are human too. I find value through AI research. If AI research is banned, what am I going to do afterwards? Why don't you support a radical platform that benefits us both instead of an anti-scientific policy that will ruin my life?

I want you to find fulfillment through art, but if art takes precedence over AI, then what do I get out of that? I like AI algorithms better than art. Their existence doesn't negate human interests under the right political system. (TBH, I find your level of paranoia dubious even under the current system.)

rotting bones wrote: ↑Thu Feb 22, 2024 5:07 pm If radical political change is impossible, then what are you even discussing? Whatever will happen will just happen, and that's what will happen. The system cannot be changed, generative AI research will happen, and jobs will be lost. Period.

As DIzzy Dean said, predictions are hard, especially about the future. If no regulations are made, then sure, AI will continue. But that doesn't mean it'll continue just as the techbros want.

One, everyone can shove AI into their products right now because the price is artificially lowered by venture capital. It's the first idea of any dope supplier: first dose is free. Is AI quite so tenable if it has to be priced according to what it actually costs? The electricity and material costs are immense.

Two, what happens when AI companies inevitably start the enshittification process? The VC guys will eventually start demanding that AI makes money. They will have to raise prices, or find a way to serve up ads, or something.

Three, cases like the one alice brought up are the sort of thing that gets the attention of corporate lawyers. The big problem with LLMs is not that they can replace humans, but that CEOs think they can. This legal setback is a reminder that they're wrong. We're not at the point where the suits start to get skeptical of AI, but we're also just beginning the stories of AI foolishness that can hurt their bottom lines.

And four, fads don't last forever. An example might be superhero movies, which took over the blockbuster slots for years, but seem to have tanked last year. This doesn't mean AI will disappear, but neither the hype nor the catastrophes are likely to occur.

For me to side against AI research, I would have to value human art more than AI research, and I don't.

Just for the record, I think that's a very weird preference.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm As DIzzy Dean said, predictions are hard, especially about the future. If no regulations are made, then sure, AI will continue. But that doesn't mean it'll continue just as the techbros want.

Again, I support regulations on an AI's dataset. As an AI researcher, I support the government making my job harder for reasons unrelated to my personal preferences.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm One, everyone can shove AI into their products right now because the price is artificially lowered by venture capital. It's the first idea of any dope supplier: first dose is free. Is AI quite so tenable if it has to be priced according to what it actually costs? The electricity and material costs are immense.

They are already slimming down the models.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm Two, what happens when AI companies inevitably start the enshittification process? The VC guys will eventually start demanding that AI makes money. They will have to raise prices, or find a way to serve up ads, or something.

Three, cases like the one alice brought up are the sort of thing that gets the attention of corporate lawyers. The big problem with LLMs is not that they can replace humans, but that CEOs think they can. This legal setback is a reminder that they're wrong. We're not at the point where the suits start to get skeptical of AI, but we're also just beginning the stories of AI foolishness that can hurt their bottom lines.

I think AI research should be a government job too. I'm not into AI research because it's profitable. I genuinely like doing it.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm And four, fads don't last forever. An example might be superhero movies, which took over the blockbuster slots for years, but seem to have tanked last year. This doesn't mean AI will disappear, but neither the hype nor the catastrophes are likely to occur.

There have been many AI winters. On the other hand, mathematical ML theorists (e.g. Anirbit Mukherjee at Manchester University) think only AI will survive and human coding will be a passing fad.

Without the profit motive, our fads can live as long as we do.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm Just for the record, I think that's a very weird preference.

This is a conlanging forum.

rotting bones wrote: ↑Thu Feb 22, 2024 5:51 pmAI researchers are human too. I find value through AI research. If AI research is banned, what am I going to do afterwards? Why don't you support a radical platform that benefits us both instead of an anti-scientific policy that will ruin my life?

You can always find other fields within computer programming. The tech industry was massive even before the AI boom over the past few years. Meanwhile generative AI has put the entire art world in jeopardy. Artists can hardly switch from manga-style portraits to psychedelic fantasy or abstract expressionism when AI has mastered all those styles and more. It might sound harsh, but you really must consider the negative impact of your field on humanity at large and ask yourself whether that impact is worth your personal enjoyment.

Quite frankly, some forms of technology are inherently corrosive to society or even threatening to humanity in general and need regulation or even prohibition for the greater good. Plenty of pharmacologists undoubtedly find their work fulfilling and enjoyable, but that doesn't make fentanyl any less dangerous. Firearms have an incredibly passionate community of enthusiasts but they also face strict regulations or bans in most countries because they are so dangerous if misused. Nuclear physicists worked incredibly hard to split the atom and undoubtedly took great pride in mastering the esoteric mathematics and physics involved. Yet nuclear weapons have multiple international treaties and agreements restricting their development and usage in various ways with many calling for their abolition altogether.

I want you to find fulfillment through art, but if art takes precedence over AI, then what do I get out of that? I like AI algorithms better than art. Their existence doesn't negate human interests under the right political system. (TBH, I find your level of paranoia dubious even under the current system.)

The continued development of AI does negate human interests because AI by its very nature attempts to fill the same niche as humans: the brains of civilization. The more powerful it becomes, the less opportunities humans have for intellectual work like literature, science, and administration. Humans are already finding themselves pushed out of art and literature en masse. Soon enough improved AI models will force everyone from physicists to physicians out of their fields. Eventually even AI researchers themselves will become obsolete once AI becomes intelligent enough to conduct research on itself. If you want to continue researching AI, then paradoxically you have a vested interest in keeping it from advancing too far.

This is not simply about art taking precedence over AI research but essentially every field of human activity based around intelligence and even the very notion of humans as intelligent beings using their minds to create their own world. We are talking about tens of millions of people in art and literature, science and mathematics, medicine and pharmacology, engineering and technology, politics and law, and countless other fields versus mere thousands of AI researchers. Your own field is attempting to expel humans from all those other fields, radically changing the nature of civilization and excluding humans from meaningful power over its evolution and activities. All this without giving us the opportunity to vote AI researchers into the incredible power they've amassed and less still to hold them accountable.

Alternative political arrangements can certainly mitigate the problem of AI displacing humans, but AI researchers probably wouldn't like the measures involved. One can imagine something analogous to the handling of nuclear materials or narcotics with AI research requiring special licenses and clearances. Universities and research institutes might have limits on how many AIs they can employ or legal requirements for every publication to have a human as the lead author. You could continue researching AI but you would have to renew your license every year and pass regular inspections. Furthermore you would watch with frustration as human academics and artists get showered with resources to carry out work your AIs could perform far more effectively if only unfettered.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm You can always find other fields within computer programming.

Are you kidding me? The tech industry has been shrinking for some time now. Four people have approached me asking if I know of job offers in recent months. This is one of the reasons ML theorists are saying that human programmers were a passing fad.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm The tech industry was massive even before the AI boom over the past few years. Meanwhile generative AI has put the entire art world in jeopardy. Artists can hardly switch from manga-style portraits to psychedelic fantasy or abstract expressionism when AI has mastered all those styles and more.
...
The continued development of AI does negate human interests because AI by its very nature attempts to fill the same niche as humans: the brains of civilization. The more powerful it becomes, the less opportunities humans have for intellectual work like literature, science, and administration.
...

You're still thinking like a capitalist. Why should human artists be in competition with AI? Humans can continue doing what they have always done, and everyone can choose which product to use.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm It might sound harsh, but you really must consider the negative impact of your field on humanity at large and ask yourself whether that impact is worth your personal enjoyment.

The 21st century "left" commonly suffers from the misconception that humans do things to feel like big shots. In reality, most of us are scrambling in the dirt, trying to survive. People support science because they expect it to get them more of the things they want (e.g. illustrations), not because science lets some of us prance around in labcoats feeling like geniuses.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm Quite frankly, some forms of technology are inherently corrosive to society or even threatening to humanity in general and need regulation or even prohibition for the greater good. Plenty of pharmacologists undoubtedly find their work fulfilling and enjoyable, but that doesn't make fentanyl any less dangerous. Firearms have an incredibly passionate community of enthusiasts but they also face strict regulations or bans in most countries because they are so dangerous if misused. Nuclear physicists worked incredibly hard to split the atom and undoubtedly took great pride in mastering the esoteric mathematics and physics involved. Yet nuclear weapons have multiple international treaties and agreements restricting their development and usage in various ways with many calling for their abolition altogether.

But research continues in all of these fields. Creating an AI algorithm is not the same as selling opioids. The analogous act would be to train a malicious AI and unleash it on someone, which should obviously be illegal. (The movie where the harasser is an ML model hasn't been made yet, right?)

Even if the Catholic Church was right about illusions being necessary for their fascist notion of social cohesion, they couldn't stop stop scientific progress. If the US doesn't fund AI research, China will.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm Your own field is attempting to expel humans from all those other fields, radically changing the nature of civilization and excluding humans from meaningful power over its evolution and activities.

This phenomenon is called Capitalism, not AI. Seriously, read Marx.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm All this without giving us the opportunity to vote AI researchers into the incredible power they've amassed and less still to hold them accountable.

Nuclear researchers are not elected. Nor are they particularly powerful except as pawns.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm Alternative political arrangements can certainly mitigate the problem of AI displacing humans, but AI researchers probably wouldn't like the measures involved. One can imagine something analogous to the handling of nuclear materials or narcotics with AI research requiring special licenses and clearances.

Why not? Nuclear researchers often support disarmament.

malloc wrote: ↑Thu Feb 22, 2024 7:16 pm Universities and research institutes might have limits on how many AIs they can employ or legal requirements for every publication to have a human as the lead author. You could continue researching AI but you would have to renew your license every year and pass regular inspections. Furthermore you would watch with frustration as human academics and artists get showered with resources to carry out work your AIs could perform far more effectively if only unfettered.

None of these measures address the actual problem. Once AI is used to deal with scarcity, of course far less efficient humans ought to be supported in the work they find fulfilling. That was always the point.

rotting bones wrote: ↑Fri Feb 23, 2024 9:44 pmAre you kidding me? The tech industry has been shrinking for some time now. Four people have approached me asking if I know of job offers in recent months. This is one of the reasons ML theorists are saying that human programmers were a passing fad.

Quite honestly, it seems difficult to take this concern seriously when you are one of the people working to make human programmers a passing fad. If you are genuinely worried about maintaining your place in an ever-shrinking field, perhaps you should reconsider your commitment to developing the forces shrinking that field. Artificial intelligence poses far more of a threat to your employment than one powerless Luddite.

You're still thinking like a capitalist. Why should human artists be in competition with AI? Humans can continue doing what they have always done, and everyone can choose which product to use.

Forget about economics then and consider ecology. When two species fill the same niche, they inevitably come into conflict. The species that better fills that niche drives the other to extinction or else out of the niche. Artificial intelligence is like an invasive species against which humans cannot compete because we never had to compete against something even smarter and more rational than us with limitless endurance and no need for food. You can call it a skill issue if you prefer and insist that humans should somehow learn to work 24/7, never making lapses of judgment or forgetting. But that doesn't change the practical reality that AI is something deliberately created and unleashed upon society with the knowledge that it will destroy livelihoods and disempower humanity.

zompist wrote: ↑Thu Feb 22, 2024 6:08 pm

For me to side against AI research, I would have to value human art more than AI research, and I don't.

Just for the record, I think that's a very weird preference.

Quoted for truth.

malloc wrote: ↑Fri Feb 23, 2024 11:11 pm Quite honestly, it seems difficult to take this concern seriously when you are one of the people working to make human programmers a passing fad. If you are genuinely worried about maintaining your place in an ever-shrinking field, perhaps you should reconsider your commitment to developing the forces shrinking that field. Artificial intelligence poses far more of a threat to your employment than one powerless Luddite.

And you think that if I personally stop researching AI, then the jobs will stop going away? You don't think, say, that if I stop researching AI, then the jobs will keep going away, and I won't be able to find a job either?

malloc wrote: ↑Fri Feb 23, 2024 11:11 pm Forget about economics then and consider ecology. When two species fill the same niche, they inevitably come into conflict. The species that better fills that niche drives the other to extinction or else out of the niche. Artificial intelligence is like an invasive species against which humans cannot compete because we never had to compete against something even smarter and more rational than us with limitless endurance and no need for food.

This is capitalists reading their anti-human economic theory back into aspects of the natural world. You can also find aspects of the natural world that are similar to socialism. E.g. The sun gives energy and asks for nothing in return.

malloc wrote: ↑Fri Feb 23, 2024 11:11 pm You can call it a skill issue if you prefer and insist that humans should somehow learn to work 24/7, never making lapses of judgment or forgetting. But that doesn't change the practical reality that AI is something deliberately created and unleashed upon society with the knowledge that it will destroy livelihoods and disempower humanity.

Once AI is used to deal with scarcity, far less efficient humans ought to be supported in the work they find fulfilling. That was always the point.

malloc wrote: ↑Fri Feb 23, 2024 11:11 pm Quoted for truth.

Discussions like this reinforce my belief that no smart human in the 21st century is constitutionally capable of believing a truth. The only way to reacquire the power to believe truths might be to read Karl Marx.

malloc wrote: ↑Fri Feb 23, 2024 11:11 pm Forget about economics then and consider ecology. When two species fill the same niche, they inevitably come into conflict. The species that better fills that niche drives the other to extinction or else out of the niche. Artificial intelligence is like an invasive species against which humans cannot compete because we never had to compete against something even smarter and more rational than us with limitless endurance and no need for food. You can call it a skill issue if you prefer and insist that humans should somehow learn to work 24/7, never making lapses of judgment or forgetting. But that doesn't change the practical reality that AI is something deliberately created and unleashed upon society with the knowledge that it will destroy livelihoods and disempower humanity.

The key thing is that AI requires huge amounts of power and huge amounts of cooling, so once the AI companies run out of venture capital money to burn and have to really sell their AI's I'd expect costs to go up considerably, such that real live humans are actually competitive.

Travis B. wrote: ↑Sat Feb 24, 2024 9:59 amThe key thing is that AI requires huge amounts of power and huge amounts of cooling, so once the AI companies run out of venture capital money to burn and have to really sell their AI's I'd expect costs to go up considerably, such that real live humans are actually competitive.

We can only hope so. Though those venture capitalists have pockets like the Mariana Trench so it could take a while. They could always find a breakthrough that makes AI much more efficient as well.

rotting bones wrote: ↑Fri Feb 23, 2024 11:29 pmAnd you think that if I personally stop researching AI, then the jobs will stop going away? You don't think, say, that if I stop researching AI, then the jobs will keep going away, and I won't be able to find a job either?

Obviously not, but you must admit that you are contributing to the very problem you lament. If someone complains about smog hurting their lungs while chain-smoking, they inevitably sound insincere. Individual choice will not solve the problem of dangerous technology proliferating of course. My choice to avoid firearms does nothing to change the state of gun manufacturing or the culture surrounding guns. Even so, I would have no right to lament mass shootings or wartime casualties if I spent all day designing ever more powerful guns and defending the gun industry against critics. Clearly we need some form of collective and indeed international action on this issue, much like previous attempts to regulate nuclear weapons and address global warming.

This is capitalists reading their anti-human economic theory back into aspects of the natural world. You can also find aspects of the natural world that are similar to socialism. E.g. The sun gives energy and asks for nothing in return.

On the contrary, the natural world is astonishingly cruel in numerous different ways. Many animals put no effort into parental care and will even eat their own hatchlings. Others have draconian social hierarchies like chickens with their pecking order or the lobsters that Jordan Peterson so admires, and the less said about ducks, the better. You don't see skinks opening orphanages nor crows holding elections. Many technologies were developed precisely to shield us from the unquenchable cruelty of nature: houses and clothes for protection from the elements, antibiotics for protection from germs, and so forth. Let us not make the mistake of creating machines capable of treating us the way invasive species treat everything in their new stomping ground.

Once AI is used to deal with scarcity, far less efficient humans ought to be supported in the work they find fulfilling. That was always the point.

The point for most AI researchers has always been replacing humans (whether physically or conceptually) with superintelligent machines. They dream of unleashing what they call the singularity, where AI becomes so powerful that humanity can no longer control or even comprehend it. The rest are staunch capitalists who want to destroy even more job opportunities than previous forms of automation so they can slash labor costs ever further. Socialism is quite a minority position in the tech industry, even compared with other industries in this era of aggressive neo-liberalism.

Putting aside everyone from Musk to Moldbug, it seems less than obvious that superintelligent blackboxes would solve the problem of scarcity. There are many factors behind scarcity, from physical shortage of resources to markets failing to make goods affordable to the poor. Technological innovation has its place in addressing these issues since it can allow us to access more resources or produce enough for everyone. Yet it cannot solve them on its own and indeed not all technologies are appropriate for addressing scarcity. In any case, most usages I have seen proposed for artificial intelligence have nothing to do with relieving scarcity. Nobody is starving to death because they can't get more anime-style pinups or fanfiction than human content creators can currently provide.

Discussions like this reinforce my belief that no smart human in the 21st century is constitutionally capable of believing a truth. The only way to reacquire the power to believe truths might be to read Karl Marx.

Marx has much to admire and made many great contributions to economics and political thought, but he also got many things wrong and had many limitations on his knowledge simply because of when he lived. Stalin not only read Marx and built his entire career around implementing his ideas but that hardly prevented him from making numerous irrational and outrageous decisions.

Travis B. wrote: ↑Sat Feb 24, 2024 9:59 am The key thing is that AI requires huge amounts of power and huge amounts of cooling, so once the AI companies run out of venture capital money to burn and have to really sell their AI's I'd expect costs to go up considerably, such that real live humans are actually competitive.

A large part of ML models is wasted computation. There is research into doing them considerably more cheaply. There have also been implementations of image detection neural nets made of biological neurons, and those use less energy.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm Obviously not, but you must admit that you are contributing to the very problem you lament. If someone complains about smog hurting their lungs while chain-smoking, they inevitably sound insincere. Individual choice will not solve the problem of dangerous technology proliferating of course.

I'm shocked that people find this surprising. Capitalism has always done these things you complain about, only more slowly. Big business is always putting themselves (in addition to small businesses) out of business. Marx calls this the immanent "contradiction" of capitalism. Everyone (including business leaders) is caught in a cycle that only a few celebrate.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm My choice to avoid firearms does nothing to change the state of gun manufacturing or the culture surrounding guns. Even so, I would have no right to lament mass shootings or wartime casualties if I spent all day designing ever more powerful guns and defending the gun industry against critics.

Beau of the Fifth Column is an anarchist firearms instructor who laments mass shootings all the time. I'd argue that situation is more analogous.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm Clearly we need some form of collective and indeed international action on this issue, much like previous attempts to regulate nuclear weapons and address global warming.

That's what I'm saying, but your choice for collective action is reactionary in the sense that it increases collective scarcity.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm On the contrary, the natural world is astonishingly cruel in numerous different ways. Many animals put no effort into parental care and will even eat their own hatchlings. Others have draconian social hierarchies like chickens with their pecking order or the lobsters that Jordan Peterson so admires, and the less said about ducks, the better.

Jordan Peterson is wrong about lobsters. Watch Cass Eris: https://www.youtube.com/playlist?list=P ... KOxtFhDZdF

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm You don't see skinks opening orphanages nor crows holding elections. Many technologies were developed precisely to shield us from the unquenchable cruelty of nature: houses and clothes for protection from the elements, antibiotics for protection from germs, and so forth.

The natural world is inhuman. All you have to do to create a society where people are not forced to behave like beasts in an ecology is to create government jobs by command. (I prefer these commands to be the will of the people for both ethical and economic reasons, i.e. measuring demand.)

Personally, I think applying "ecological" thinking to human society is an essentially conservative policy. I oppose all popular political factions because it's a contest between the "ecological" conservatives (left) vs. the fascists (right). Neither of these groups are looking out for my interests.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm Let us not make the mistake of creating machines capable of treating us the way invasive species treat everything in their new stomping ground.

AIs have no instinct for survival, and no one is incentivized to give them one. Corporations like AIs BECAUSE, unlike humans, they have no instinct for survival.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm The point for most AI researchers has always been replacing humans (whether physically or conceptually) with superintelligent machines. They dream of unleashing what they call the singularity, where AI becomes so powerful that humanity can no longer control or even comprehend it. The rest are staunch capitalists who want to destroy even more job opportunities than previous forms of automation so they can slash labor costs ever further. Socialism is quite a minority position in the tech industry, even compared with other industries in this era of aggressive neo-liberalism.

Putting aside everyone from Musk to Moldbug,

These people are not AI researchers. Honestly, I'm not sure Elon Musk functions like a full-fledged person at this point. If you count people like this, Astral Codex Ten is a left-leaning blog that started as an AI blog.

I know that a lot of left-wing AI researchers exist. Even Anirbit Mukherjee I mentioned earlier says his politics is mostly opposing the right, but he's not left-wing either. Personally though, only right-wing sources have ever offered me opportunities to grow my career. Throughout my life, I have never even had a left-wing professor.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm it seems less than obvious that superintelligent blackboxes would solve the problem of scarcity. There are many factors behind scarcity, from physical shortage of resources to markets failing to make goods affordable to the poor. Technological innovation has its place in addressing these issues since it can allow us to access more resources or produce enough for everyone. Yet it cannot solve them on its own and indeed not all technologies are appropriate for addressing scarcity. In any case, most usages I have seen proposed for artificial intelligence have nothing to do with relieving scarcity. Nobody is starving to death because they can't get more anime-style pinups or fanfiction than human content creators can currently provide.

Scarcity of illustrators for poor writers. Once I'm fired from programming, shut up in an insane asylum and writing novels to pass the time, who will do my illustrations for me if not DALL-E?

There is also a lot of information-based scarcity that AIs solve. My research is in automated debugging. This might put human debuggers out of business, but it's good for programmers in general.

malloc wrote: ↑Sat Feb 24, 2024 12:17 pm Marx has much to admire and made many great contributions to economics and political thought, but he also got many things wrong and had many limitations on his knowledge simply because of when he lived. Stalin not only read Marx and built his entire career around implementing his ideas but that hardly prevented him from making numerous irrational and outrageous decisions.

I don't think it's true that Stalin implemented Marx accurately. An accurate implementation of the Marxist vision would have been better than Stalinism in some ways and worse in others. I have written extensive comments about my disagreements with Marx. Nevertheless, he gets many important things right that nearly everyone gets wrong these days. This makes perfect sense once you remember that accuracy is unnecessary for success.