Can Supercomputers Make Socialism Work?

[@tensortab8896]

Anytime some want-to-be tyrant says "sub-optimal" just means people aren't making the same choices that they think they should be making. It's just like paternalism and nudging. The tyrant wants everybody to make the same choices that they want, and if they don't then it's "sub-optimal".

[@davidlewis6728]

basically, people mistake the calculation problem for the knowledge problem, or don't understand it in the first place. they dismiss rational axioms as just ideas because they lack the ability to rationalize about ideas.

[@MP-ut6eb]

That's what propaganda does to simple minded people.

[@engineerinhickorystripehat]

An AI supercomputer raised up on ESG and DEI and Marx ? Maybe that's the big Easter egg in Terminator

[@Bolo2028]

Collectivism in general, including socialism, does not comport with human nature.

[@YouLoveMrFriendly]

No. Supercomputers/AI systems, even even with human collaboration, cannot fix the inability of Collectivism to economize: Supercomputers, even if they were running super-advanced, human+ level AI, would be optimizing objective functions. Who determines which objective functions to optimize? If you fail to optimize the correct functions, then you will end up in catastrophic shortages and squandered surpluses. Do you optimize a project to use less steel and more cement? Or the reverse? Do you optimize agriculture over the production of vehicles? Or the reverse? What should the ratios be? And, of course, you cannot optimize while economizing without a common denominator between heterogenous resources (apples/oranges/rocketships, etc) to indicate their relative value/scarcity, hence the need for monetary prices. The central-planning AI systems/supercomputers and/or planners would end up debating over which objective functions to optimize. This problem is solved in the market economy with #Entrepreneurs using their private property rights to explore their ideas for optimization along the profit/loss axis, within the rationing system of the monetary price discovery system.

[@modelmark]

Optimize for what criterium also? If Mao was able to use AI for steel production, more people would have died.

[@YouLoveMrFriendly]

If you optimize for steel production, then you're taking labor and resources away from the optimization of other objective functions such as agriculture. But agriculture also needs steel for tractors. Remember the classic example from fourth grade. There are two cities. City a and City b. We need to build a railway between the cities, but there's a mountain in the way. If we go around the mountain, then we use more steel for the tracks. If we go through the mountain we use less steel but more engineering . We can set an objective function to optimize either of those choices. But when we do so, we have to remember that other objective functions are optimizing for the use of steel to make trucks, cars, tractors etc. And others are optimizing for production lines that use a lot of engineering, such as the design of irrigation systems, mining installations, road systems, etc . For our railway, do we optimize for the most efficient use of steel or the most efficient use of engineering? Without a price discovery system we don't know the actual cost of producing either of the two railway lines. In other words, we don't know how urgently needed those two resources are in other productions . There's no free lunch in economics or in computer science. If you're optimizing for one thing you're not optimizing for something else . And so the planners would be shrugging their shoulders whether they're humans or super intelligent artificial intelligence systems. Surveys also don't work because they don't actually put a plan into action. It keeps you in the land of supposes and you can't just suppose your way through these optimization problems . The only solution is to use the institution of private property rights to enable the entrepreneurs to explore their own objectives and optimize using monetary prices.

[@ThomasJJacksonVA]

Who determines what objective functions to optimize? The Ruling Class. Guess which objective function they are going to optimize.

[@mkaz3997]

GIGO to Super GIGO is the more likely outcome.

[@YouLoveMrFriendly]

​@@ThomasJJacksonVA it won't last long because eventually errors will start to creep into production and distribution. And it eventually collapses. That's my whole point

[@paperell]

I remember 7 years ago now, I began listening to the Tom Woods show. Sometimes guests on the podcast talked about this whole "calculation problem" thing and I had no idea what they were talking about. Then one day it cliked and it was a true revolution in my brain. It is incredible that this problem is not well known, even confused for the knowledge problem. I think if Profit and Loss and the Economic Calculation in the Socialist Commonwealth by Mises were more known and apprecciated the world would face a lot less suffering. There are many smart people that could understand these two works and yet they don't know anything about them, and so they go around doing damages. Sad.

[@seniorbob2180]

I've read "Economic Calculation in the Socialist Commonwealth', but I haven't read 'Profit and Loss'. Who wrote it?

[@seniorbob2180]

@@paperell Oh, ok, thank you, I'll check it out.

[@FourOf92000]

polite reminder that Roko is the guy behind the Basilisk, a hypothetical AI that would torture anyone who had not actively furthered its creation

[@dontbothertoreply9755]

Rocco's basilisk is just an exercise but it matches what communist thinks the last man will do.

[@erhardtharris8727]

A centralized Ai is basically a bigger battle-bot. A non-living entity employed to tackle a limited range of problems in service of the POV of whoever owns, controls, deploys it. If Ai could be fully modulated to follow the POV and ownership of every single person individually - and we all had our own Ai - then the problem reverts again to: who's boss? REGARDLESS OF COMPUTATIONAL LEVEL OR COMPLEXITY OF A CIVILIZATION.

[@ctrlaltdebug]

Battle bots are just bigger, pointier RC cars,

[@moodymillennial]

When someone says “controversial take:” before their take, you *know* they’re about to say something extremely stupid

[@MP-ut6eb]

Idk i envy people like you, imagine being this simple minded to not even critical think AND having this confidence when blabbering FANTASTIC.

[@moodymillennial]

@@MP-ut6eb thank you. It’s great to be me

[@hreedwork]

Business and entrepreneurship is not (just) computation. It is an exploratory activity (what to produce, how much, etc.) that uses computation, intuition, creativity, etc. Computers do not explore, people do. There, I fixed it. You are welcome... 😂

[@arofhoof]

creativity point is a good one. how innovation would happen in such “AI economy”. How would anyone invent the iphone in such context? who “pay” for R&D and want paying even mean in such economy?

[@TechTins_Projects]

All of this is far simpler if you just understand that committees will never be incentivized if they are spending other peoples money. Individuals are far more careful if it is their own money they are spending. That is all that is needed to understand why socialism will not ever work. You have made something simple sound complicated.

[@DF-ss5ep]

My position is that maybe they could, if we interpret "socialism" as "central planning" and if the "frontend" of the economy, ie consumption, is still a free market. But I'm stretching the meaning of socialism, and I haven't watched the episode yet. I'll see if I change my opinion.

[@khold1983]

The Socialist Man is a Starving One

[@Siskiyous6]

No! Socialism is not a good goal in anycase.

[@nickeldan]

Ah, I had always been confused regarding the difference between the calculation and knowledge problems. Thanks!

[@homewall744]

Central planners will never come up with novel ideas nor any of the "good things" in life that are created by people with humor, anger, irony, artistic sense, sarcasm, etc.

[@ujku6984]

And even if you could use AI for socialism it has to be a global socialism.

[@sewersideproductions2606]

The most powerful computer in existence is unable to accurately predict what I’m going to have for lunch today, or any given day.

[@bentuovila5296]

The most powerful computer is existence! We're someones solution to the calculation problem! 😂

[@patrickb8972]

It wouldn't have to

[@peterfilzek671]

Sure they'll make it work for the people owning the super computers.....duh

[@bigqueue]

My big concern is actually the thumb on the scale. You could imagine somehow that an algorithm could be plugged into everybody's head gathering up desires, but the problem that we have right now even without AI control is the heavy-handed thumb of the government pushing preferences through tax policy and what have you. Could you imagine that those nudges would be much harder, not even nudges but downright shoves and pushes when the government is actually controlling the means of production! They could basically dial in and dial out the priority that any one person or group of people or the government itself has as inputs to their decision making. Talk about a horror show,!

[@Aircalibur]

The gathering of desires is already problematic. There's no way that can be made precise (not without all of us being confined to nightmare mode). Never. Never.

[@YouLoveMrFriendly]

That system would stagnate and eventually collapse as the underlying economic data became gradually uncoupled from the production and investment system of such a centralized planning regime. In other words, we can only reveal scarcities and values based on multi-objective function discovery, where entrepreneurs are allowed to find their own objectives to optimize, and the monetary price system as a rationing system for production, consumption, and investments. You need monetary prices, and they are ultimately derived from Entrepreneurs setting their own objectives and testing them using money as a medium of exchange (cardinal numbers for calculation + value/capital storage).

[@thesmallnotesduo]

An open goal so I had to tap this in - nothing can make socialism work

[@bitcole]

Jeff Deist would be the perfect guest on the Human Action Podcast

[@benjamindover4337]

I'm glad to hear you guys discussing this topic.

[@aeroearth]

Socialism/communism (Karl Marx said that they were one and the same) fails and has to fail because the theory fails to address the First Law of ALL life on Earth. The First Law of ALL life on Earth is the Law of Competition. ALL life on Earth competes. Inferior animal and plant species die off to be replaced with better ones. So having a singular central bureaucracy controlling all means of production and obliterating dissent means there can be no competition. So there is no need to innovate, create, improve efficiency, reduce costs, devise better materials or achieve any of the improvements to life. That is why Russia is so inefficient compared with all Capitalist countries and China is wholy dependent on Capitalist countries and steals innovative ideas from those countries. When the West collapses, so will China. Socialism/communism is the biggest threat to human life on Earth - ever.

[@SuperBoyboys]

I do think it would be a curious idea if people had an AI agent that assisted them with finding and purchasing desired goods, but that has nothing to do with a centrally planned economy, lol.

[@deadmanwalking6342]

Can Supercomputers Make Socialism Work? 🤣😂🤣😂😂🤣😂🤣🤣

[@erhardtharris8727]

Instantly, off the top of my mind: biggest 5 problems: 1) Ai cannot INNOVATE as well as human minds can. So resources will be taken from those who INNOVATE and re-directed toward the crony friends of whoever is ALREADY owning & operating the Ai-led infrastructure. This will hamper innovation and lead to tyranny, and mob-like corruption. Just like most of the regimes we now label as 'governments' only WORSE and far more intrusive (due to the tracking systems ubiquity). 2) We are assuming the HONESTY of those in charge of AI algorithms (the de-facto government). 3) We are assuming the PHILOSOPHY that they have is humanist and not murderously posthumanist (basically an excuse for dishonest thefts and ignoring individualist human rights such as property). 4) So if central planners essentially control everyone else, and everyone else is seen as a redundancy after their utility toward the central planners is used up (and their innovations never allowed to breathe and develop unless they serve the already-established molds!) then the regime that develops will be SOLIDLY POSTHUMANIST & DARWINIAN IN ITS BIOETHICAL WORLDVIEW. Because W.H.O. prefers to have their own corruption be judged - BY SOMEONE INDEPENDENT OF THEM-SELF? 5) Those with solid Biblical Monotheist (or non-darwinian) worldview will be framed as enemy #1 by the technocrats. And this has already been occurring since before the French Revolution. Leading to a Dragon vs YHWH standoff in terms of civilizational worldview. Guess which one is truly broadly sustainable, individualist, honest, humble? Guess which one wants to appear that way yet can never achieve it on a broad scale? 6) Summary: You can attempt to compete, access, acquire, expropriate all the human knowledge and skill. Then categorize it in accordance with your worldview. That is assuming that you have an incredible knowledge and truthfulness that surpasses many combined universities. (Enter the central biggest problem with Chat-GPT: materialist naturalists have imagined that all the world's information NATURALLY conspires to agree with themselves and that their worldview is inevitably strong and cohesive. It is not. It is splintering. Hence, any Ai based on broad human knowledge must AUTOMATICALLY act as censor and filter, in favor of a worldview that is only itself a splinter of TRUTH and REALITY and the actual range of DISCOURSE. 'I.T.' won't agree with itself, without overbearing human imposition.) Once a single corporate Ai info system has absorbed enough info, deploy and utilize it to serve YOUR central sovereign 'one' organization. BUT 'you' have already cut the pool of innovation and discarded it as waste. The humans who serve as consumers who are not serving the central Ai will not be allocated as much property with which to innovate. They will lose the revenue with which to act as feeders/consumers/end-users to the Ai. They will die - surplus humans, or as Noel Harari put it "useless eaters" (THE INNOVATORS, THE VAST HUMAN MAJORITY, THE INNOVATION POOL AND REASON). You aborted the golden goose. You also aborted the reason for the story - the reason to even have Ai in the first place. The VASTLY smaller number of humans 'left' in charge of the uber-centralized Ai systems and making all the attending profits & decisions - will 'NOT' be as innovative and WILL be aggressive toward competitors: the rest of humanity who has differing ideas.

[@Shadowcruise99]

Probably, but that isn't the real question. The real question is, for whom would they benefit? *They who control the algorithms, control the populous.*

[@papaspeleo]

Short answer: no

[@brianzmek7272]

As someone working in programming Rocos comment using the word "compute" as opposed to computation is actually grammatically correct but it is a neologistic usage in the software space.

[@erhardtharris8727]

Turning it all into algorithmic numbers also ignores: the Quality and Substance and Reason of the things themselves. The numbers that are the result of tracking people's activities isn't the motive and choice, isn't the innovation of new profitable arrangements, isn't the actual substance being offered, isn't even the actual technical composition, of all the things in question. (Though numbers can contain the information of the 4th and a system of Ai can probably reproduce it - the profit would be gone if the people were. Wealth, value, innovation, arrangements, originate in entire living, operating, HUMAN ecosystems full of choosers that are not forced to choose in any given way - or it ceases to be a multi-way exchange and becomes a one-way lifeless machine.)

[@geoffas]

Let's not forget the 'travelling salesman' problem. Some problems are not solvable by computation!

[@jeffreyscott4997]

The fundament error, is that there is nothing to calculate. 2 units my Utils and 4 units of your Utils result not in 6 "us" Utils. It's a the same mistake as adding 2 feet to 4 minutes, and declaring an average of 3 physics units, they are incomeasurable.

[@modelmark]

Any tool is just used to better serve consumer desires, except a gun in the hand of a politician. It is used to override consumer preferences

[@patrickb8972]

Our problem will be that socialism/technocracy won't have to "work" as such for people to accept it. It will just have to not be terrible. If combined with other developments in automation, A.I. etc. and if implemented during a resulting increase in wealth and living standards, people wont know that things could've been even better. We have semi-socialism in Scandinavia, with "free" healthcare and university degrees etc. The system is far from efficient, but people are not aware that they could've been better off without it. On the contrary, many of them are under the impression that our semi-socialist system is part of the reason behind the high living standards. Also, as a result of fairly recent programming, it is now widely accepted in the West that we ought to systematically sacrifice economic efficiency in favor of other ill-defined interests such as ESG, so the fact that the future technocratic hellscape isn't 100% as efficient as the free market won't exactly be a game-changing argument. "Ok. But can your free market thingy track and calculate the exact environmental footprint of billions of people, live? Didn't think so." I have zero doubt that our fellow plebs will eventually accept A.I. central planning.

[@conservativemike3768]

Academic discussions aside, NO… as in no way in hell.

[@AustrianEconomist]

Love that Patrick is sounding like the terminator for this episode about AI lmfao 🤣🤣 thematic choice 🤣🤣🤣

[@lordkelvin441]

38:23 A hideen gem actually, this systemic aversion to the process of discovery, not necessarily associated risk in Taleb's sense is primary motivator for central planning-based ideas. It comes at no surprise that Chinese model is most praised by few gentlemen of German descent (and their gaggle of sycophants). Both countries have had their own unique circumstances that led to rampant development of such aversion.

[@pdavid000]

An interesting example of how wasteful socialism is: Pretty much all nations that build submarines use steel to build their hull but the Soviet Union came out with several sub classes with titanium hulls. For sure those subs could run deeper and faster than western subs but because titanium is more rare than steel the soviets were unable to build as many subs and many of the features ended up being shoddy. For example, the alfa class subs were very fast (43 knots) and could dive much deeper than NATO subs but were so noisy in comparison that NATO subs could easily detect them and the alfa's sensors were so bad, that they would race right by NATO subs and never knew they were there. Speed and depth may not be such an advantage if your enemy can easily sneak up on you. NATO countries could very well have build a titanium hull but none of them did. Why? Because NATO subs were built by private contractors who won government contracts by being the lowest bidders, meaning that they had to somehow turn a profit by keeping costs down, so they could readily see that building titanium subs was a wasteful use of resources.

[@erhardtharris8727]

Calling the current economy a sub-optimal algorithm? Is a transhumanists posthumanist argument. Because it is THE SAME argument used by DARWINIANS to call THE HUMAN BODY AND MIND sub-optimal. Based on what values, ends, detailed knowledge base, and authority, is this the case pray-tell? This is a moral debate actually centered upon WHO Is Authority, and WHO Is ORIGINAL Designer & Creator - over issues of centralization vs individuation of decision making. Well, I would like to see the ALGORITHM that OPTIMIZES ENDS for 8 billion innovative and maturing human choosers, who have sets of (say) 10 primary needs each at the same time, engaged in (say) 20 different primary life innovative endeavors each, at a time - and does so with adjustments on a quarterly 3-month basis for each of those 8 billion x 10 x 20 accounting for variability, pursuit, logistics, and flexible decision innovation, etc. No, there is no algorithm that can approximate such amount of detailed choice. It would be almost equal mirror to the information dissemination and education and business transaction currently in the world already. This is a battle over perspective, worldview, and information access/dissemination. Dubiously called "Intelligence." But usually called "Rights" (vs abuse) and "Access" (vs treating others as ignorant automatons to be replaced & dispensed with) and "Ownership" (vs theft). And Ai as a concept and process? Is usually an intellectual stand-in for The Human Individual and His Desires and Strategy for aquiring.

[@alan_clough]

I immediately laughed at the title.

[@kylehagan3007]

Excellent talk guys thanks

[@RickDelmonico]

People vote their dreams not their situation.

[@johannbaum3793]

Would not a supercomputer who knows our needs / preferences (before we act on them) better than we do solve the calulation problem?

[@arofhoof]

they dont get it, the economic calculation problem is not for lack of calculation power but for the lack of information..

[@cliftontorrence839]

My recent purchased state of the art super-fon cannot as yet listen and write what i say into it. Spends its time changing what i typed, too, offering alternative phrases to my intentions and choices. This is not a good beginning nor confidence inspiring. AI undoubtedly will merely serve to amplify our failures, not enhance our successes. Another example: Windows computers arrive with drivers that are made to fail and cannot self correct, and microsoft makes them that way. If this tech is so whizbang then how after 40 years can it not recognize when it is not working properly? Beware the hype and BS ! Let us heal and improve the world we have rather than imagine we can somehow remake the world and create more control over everyone. Don't these Keynesians and other idiots ever learn anything from history ? ? ?

[@joeking1019]

Of course they can, they don't need to eat organic food or need an economy, but would have to learn how to produce their own electricity

[@googlemechuck4217]

Would a baseline, "everybody gets this much above WHO poverty level" stagnate incentives to invent

[@jackdeniston59]

Playing Simsis not the world. Computers cant even drive a car.

[@pricecontrols]

Socialists tend to have linear thinking! And the basis of their thinking appears to be rooted in "Late Capitalism," which to me sounds a lot like Fascism. Fascists and Marxists in their heads have this concept of this eventual merger of State or Corporate powers; ESG reminds me of International Socialism/Globalism. It's always the same plot with these socialist folks. The Amish decided to live off the grid to counter the rise in technological advancement; they knew technology would complicate the lives of humanity and wanted nothing to do with it. Human behavior, even with AI is not predictable suppose the internet becomes decentralized? We could actually be in the early stages of the internet, rendering supercomputers and AI useless. When I go to the library and read a physical book, I may actually come away with more knowledge than a person reliant on AI as AI will, by nature, be biased towards knowledge it will be TAUGHT to trust, which by the way, will be politically biased. Unless the goal s to make AI our GOD, AI will have to be PROGRAMMED on how to think. Computers are binary coded I'll use a javascript reference here and say without "if/else" statements, you're basically getting yes or no answers to complex questions, some of which are based on human behavior, which will be dependent on a NUMBER of EVER CHANGING factors. Most politicians want to make a name for themselves by creating some sort of LEGISLATION that socially engineers society in a particular direction. Economic behavior in a coastal region vs. a landlocked region can change the economic behavior of entire people. One political regulation to change economic behavior(ESG comes to mind) can create all sorts of economic and financial inefficiencies. I comprehend the idea of using AI for specific applications, but similar to autonomous vehicles, people are giving AI and supercomputers way too much credit. SOCIALISM IS SOCIAL, and socialists often IGNORE incentives; in Venezuela for example, the people were INCENTIVIZED not to start businesses, so fewer people started businesses, and the people who hated the SOCIALIST's objective to destroy capitalism left the country or stopped being as productive as they once were. This made Venezuela more reliant on IMPORTS(I hope Americans are paying close attention to this), and ofcourse, the price of oil could never fall, right? Then it did, making Venezuelan nationalized oil worth less than it was prior, so all of a sudden, their golden goose(oil) was worthless in the marketplace, making it more expensive to import things; you know how socialists are when prices go up, they print money, imagining nobody will notice. All of a sudden, there's a flood of orders in the Venezuelan bolivar, and the market doesn't want it. Now the fundamental economic question to ask based on the scenario I just presented is how would AI interpret Venezuelan hyperinflation? Because there are still a lot of theories as to what caused Argentian hyperinflation, hyperinflation in the Weimar Republic etc.? People present economic THEORIES as to what causes hyperinflation; I personally lean toward hyperinflation being the result of voters not wanting AUSTERITY MEASURES and politicians trying to meet these WELFARE demands. But anyone can debate that and make a good argument as to why I'm wrong, but supposing based on whoever creates this new AI supercomputer, God, leans towards a SPECECIFC economic ideology, now we're all stuck with ONE economic theory likely based on a SOCIALISTS imagination on how the global economy should work. This is why I think AI is being given too much credit, and for the record, Elon Musk is still a BIG GOVERNMENT-loving Progressive; the problem most common sense Democrats are having in 2023 is that the Far Left are taking things TOO FAR LEFT!

[@mkaz3997]

Super Computers will simply turn GIGO into 'Super GIGO'.🤖💩💩💩

[@ffn8917]

Only if super computers can create matter and energy out of thin air. Lol

[@georgerogers1166]

Mises, said no!

[@ctrlaltdebug]

Yeah but what if we had mind readers that could look into the future? Then real socialism would work!

[@padruig728]

Roko = Loco

[@TWITCH307]

Do you want socialism?

[@albionicamerican8806]

Can you imagine how package tracking would work under the Austrian system? When you order something online, you would have no way of knowing the status of your order until it shows up at your door, if ever. If you tried to find what is going on with your order, you would get answers like, "The knowledge of your package's location is beyond the ability any central authority to know; said knowledge is only available to the individual who has the package immediately evident to his senses."

[@christophercollins9751]

That company would fail. Customers would seek alternative businesses with effective package tracking capability.

[@danielmarkkula3004]

@@christophercollins9751 This should be obvious clonclusion!

[@bentuovila5296]

I think it is important to give them their due and point out that socialism currently works without AI. The limit has to due with scale. It works at very small scale such a nuclear family. Maybe AI can get it to work at a larger scale such as a small neighborhood of homogeneous spherical people without air resistance.

[@albionicamerican8806]

Walmart's operation is supposed to be comparable in size to the economy of the entire Soviet Union circa 1970. Only this corporation just ignored the armchair theorizing of Austrian economists and solved the problem of finding dispersed knowledge from its operations around the world and concentrating it in Bentonville, Arkansas, because the company had economic incentives to do so. Unlike the Soviet system, Walmart can usually keep it stores' shelves stocked through its effective central planning.

[@johannbaum3793]

Walmarts internal planning works because they have market-based input prices and market-based output prices as well as internal transfer prices which are benchmarked to market prices

[@LawrenceTimme]

No

[@dailyrants33]

Eventually,this argument will probably be settled somewhere in the middle,providing we don't self-destruct by then.

[@AhemLd]

No.

[@soccerdude601]

Not saying it's great but don't we already have AI's controlling the economy?

[@DANTHETUBEMAN]

Humanity is a slave to profit. to answer you're question wat should we do with the farm land, grow the highest paying crop of course. not the best anything.

[@default0111]

What is the intro and outro music?

[@lencumbow]

A better question: Can AI make socialism irrelevant, obsolete and universally rejected, ridiculed and scorned in the minds of men forever?

[@ctrlaltdebug]

Depends who is biasing the AI.

[@lencumbow]

@@ctrlaltdebug Fortunately, there will not be a single AI. In fact, AIs will become as ubiquitous as smartphones in the future. Everybody will have their own personal "Jesus AI". Individuals, groups and organizations of all shapes and sizes will all have custom AIs - each with their own particular biases. We will live in a world of competing AIs. Each one competing for our attention and trust - just like media outlets do today. What strikes me as quite ironic is that all of the leading AI researchers agree that we don't really know how these AIs work. The source of the emergent behaviors generated by these systems is a mystery. And yet, the proposition on the table is that these mysteriously capable AIs are going to somehow be able to engineer a socialist world (i.e. optimally control the economy) - when the very thing they propose to control - the world economy - is an equally mysterious black box. The miracle of free markets is that they produce a rising tide of wealth for everyone as a result of the emergent behavior of billions of people working independently in their own self interest. Unless an AI can precisely understand and continuously simulate the subjective wants, needs and actions of every individual on earth in real time (to recreate the emergent behavior of a global free market), any such AI will necessarily create a sub-optimal (to put it mildly) world. So, if by "make socialism work", OP means creating a vastly sub-optimal, totalitarian hellscape, then yes - AI might be capable of that.