AI Is Useful for Capitalists but Probably Terrible for Anyone Else
mebassett.info
mebassett.info
Once those laws are in place, the moat is dug - and large corporations with access to capital vacuum up good training data in bulk. The whole time loudly proclaiming that they're 'empowering creators' and 'rewarding original thinkers.' Now the large corporations have the largest and best datasets tied up and neat, deep moat. Any startup wanting to challenge the incumbents could have significantly better models, but without access to the same quality and volume of training data, they'll be at a massive disadvantage.
The law is simple... You have to have a license to the data to train a model with that data. If you don't... Get ready to be sued.
There's no good way to put the AI genie back in the bottle, or make sure it will not create catastrophic undesired consequences. Only way is forward, even if there is no good path in sight.
This won't be the case everywhere. UK has actually already explicitly ruled the opposite
any law can be legislated, if it can be put down as text.
It may be considered unreasonable, but if the majority of people _do_ agree, then it could be done.
To operate in the real world like humans you need to have motivation. What is an AI motivated by?
There is little to no need for anyone to scrape the web themselves, CommonCrawl does it for free and provides access to all that data for free, they are a charity similar to The Internet Archive:
As for using that data to train a model, I've done it at my own company and it's not that expensive, certainly not the astronomically expensive figures people seem to think it is. Maybe a hobbyist can't do it for themselves, but my small company managed to get pretty good results training on our existing hardware. It's hard to give concrete figures because we manage our own hardware instead of using AWS, but all in we're talking on the order of 100k, as opposed to what I've heard people on HN think it costs, as if it's a $10 million+ enterprise.
Remember... Bits have color and AI models are a rainbow.
There are clearnet sites which host entire movies and games and there have been since the dawn of the internet, wouldn’t you expect those would be bigger targets to go after if the powers were there? All that to say, I’m not bullish on the future global leverage of American copyright law, especially when it comes to AI models.
Case in point: GDPR, you can vacuum as much private data as you wish from non-EU citizens, the moment you touch EU-citizen data you must abide by the GDPR or you don't have access to a market of 400+ million of some of the wealthiest countries in the world.
You don't need global jurisdiction when your economical power is large enough.
The real issue for machine learning is that it lets you build systems that optimize for some metric. That's what corporations do. I've been pointing out lately that the ethical problems for AI in the near term look very much like those for corporations.
I think the future you described is possible. I also think it's possible that corporations are overpaying for data that will become stale after several years and new fresh data is generated in such great quantity/quality that access to it is not capital intensive.
There seems to be a lot of different ideas about 'quality'. A lot of people want a better-than-search tool that knows vast amounts and can be assigned tasks. I'm personally much more interested in a reasoning machine, I don't need it to include the entire contents of the New York Public Library.
Seems like we almost need pre-ai scraped datasets, almost like how it’s sometimes very useful to have metals that were created before the first atom bombs (I don’t know too much about this, just that there’s a definite land in the radiated sand around that timeframe).
But the other side of this is that large corporations will have the upper hand in AI only if the success of that AI is dependent on their other advantages (like having tons of data about all of us). The fact that model architectures are getting somewhat universalized seems to suggest that eventually there may be some highly reusable building blocks that will provide good outcomes ... when trained on enough data with enough diversity. And that data comes from everyone; if we insist on different data governance principles, we could support an ecosystem where anyone can create, train, and run their own AI services.
If dumping laundry detergent into a pile creates gold, but once the pile reaches 1 million lbs it causes the Earth to explode, you have to find a way that no one can gather such a pile. Not "democratize" the practice and letting anyone anywhere have as much detergent as they wish.
AI does not improve productivity in the technical sense. Productivity improvements are the result of using better tools. I would classify better tools as force multipliers. AI is a force replacement as far as I'm concerned. If I'm using stable diffusion, I'm not painting, I'm writing prompts and hoping I get what I want. It's as far removed from the actual practice as it can be.
If you used Photoshop to manipulate an image over using MS Paint, you wouldn't say that Photoshop made the image for you, because it didn't, you did. Grammarly doesn't write your articles for you, you do. They just maximise your efforts.
Did you dig the hole, or the backhoe you were driving?
Modern DJ decks have an auto-mixing switch. DJs still exist and use that switch when they need to take a bathroom break, etc.
Prompt-writing is to interacting with AI as the command line is to using Photoshop - an early UI to a new tech.
Forgive my ignorance but is this not the whole point of civilized society? Social safety net are what my taxes are supposed to be for isn't it?
I'm from Europe living in USA and the discussion around "What if there were some way to pool our resources to help those in need" is always ... interesting.
It's unwise for Americans to stay so mentally anchored to this origin story. It's not that it doesn't have value, but it has increasingly less to do with reality because it depends on a condition of superabundance that no longer obtains.
The personal home computer, the early web etc were productivity boosts but moved power from incumbents to small players. AI as it currently stands almost achieves the exact opposite. A relatively straight forward heuristic to ask is, how much does an authoritarian power like the technology?
Even Bitcoin is not going to fix any of this.
The claim goes too far though. Automation, thus far, has been good for everyone. Mostly those with capital, but also those without. The zero sum outcomes implied by the headline don't hold up to what we've observed with increased living standards globally as automation has improved economic output.
Once they have mass produced Terminator-like invincible robot cops they'll be able to do whatever they wish with us. They could give us universal income, they could also round us up and push us into the sea.
Being able to automate valuable work means you need a smaller coalition of power brokers to maintain the bulk of your GDP, meaning a government can remain in power with the support of an ever-shrinking class of influential people.
This is something people have been talking about for decades, with semi-serious proposals like a tax for robots being thrown around. The people in power have yet to take a hard look at any of it, though -- and why should they? It's a safer bet to curry favor with the ever-shrinking circle of influential people. The core problem in my opinion is how to create an incentive for politicians today to put into place measures to combat this consolidation of power. The longer we wait, the stronger the incentives against liberal democracy become, and the harder it will be to make a change.
They're somewhere between reactionary efforts to strangle the technology in the crib, and just completely misinformed naval-gazing by people with nothing better to do - proposing fictional scenarios to argue about rather then dealing with real data.
https://www.cbsnews.com/news/mcdonalds-automated-restaurant-...
And then even if you have the data and the engineering talent, it takes competent leadership to deploy a product and shepherd it to success. (Looking at Google’s graveyard here)
https://issuu.com/stanfordchaparral/docs/parody_119_3-4/17
To summarize: Larry Page is bragging about how, in 2017, Alphabet created a "data REIT" which contained all of Google's data, and licensed it on FRAND terms to all comers. As a "REIT" it's required to pay out 95% of its profits as dividends, and everyone whose data they use is a shareholder in the "REIT."
Yes, I know that's not what REIT's are for. This would take legislation, as Larry hints in the interview.
In this "interview" Larry is bragging about how well this worked out for Google, since the clients of Google Data can make much better use of the data than Google itself can.
This misses the deeper reality, in my view. AI is predicated on and bootstrapped by the free labor of others. Even if it were “open source” and “owned” by end-users, AI fundamentally requires people to do free work. That’s the problem with it.
― Frank Herbert, Dune
In other words, why is this time different?
The problem isn't new, it's always existed as a pathology of American society. You already have been trading away the gains of technological improvement to an ever shrinking group of people in control of capital.
AI isn't creating a new problem, it's a reminder of an old one. Which should make it clear that even bothering to talk about AI specifically is talking about the wrong issue.
i dont think that's true. The average wealth has grown, and even if you didn't get wage growth, the availability of things you could buy with said wage has grown (which is in and of itself a form of wealth).
American productivity/wage gap isn't particularly controversial, it's well documented.
>The average wealth has grown
Globally yes, but as far as Americans are concerned they largely didn't see their wealth grow as the "Elephant curve" [2] shows. A lot of the wealth growth happened in China and India.
>the availability of things you could buy with said wage has grown
"You'll own nothing. And you'll be happy" is pretty much what you are saying here. The price of real assets has grown tremendously, but at least you can buy a 60" flatscreen TV for $100.
[1] https://wtfhappenedin1971.com/
[2] https://www.aeaweb.org/research/charts/elephant-curve-world-...
The US is simply a thug. It has too much firepower for its own good. It's drunk on its ideologies of freedom or democracy or what have you, that it can never recognise its mistakes.
Singapore is a city-state. Tiny countries like Iceland or special cases like Macau or Hong Kong shouldn't count.
My point is capitalism itself is not necessarily the root of all evils; it is puritanism, dogmatism, and when it comes to the US, it's their puritanical approach to "freedom" that ironically makes the majority of the population less free and less democratic. A communist, planned economy is as much capable of creating an incompetent government and society if there were no political, ethical and cultural wisdom.
To summarise, capitalism is abhorrent, but there are other dimensions.
Nothing is the root of all evil but “evilness” has almost always been done from a right wing perspective. Yes authoritarian leftists exist but I’d argue it’s impossible not to have to behave like Cuba with freedom restrictions does when America and by extension, the Global North are capitalist and imperialist bullies coming after you.
I still mourn for my parent’s country having us have their socialist leader murdered and replaced by an awful fascistic dictator.
It kind of hurts to hear people still think this. My parents home country had coups taking out leftist socialist leaders and installing awful dictators. Capitalism is evil. Capitalists should have left socialists alone after World War 2.
I think Stable Diffusion, being FOSS & usable on a decent range of consumer hardware, is a pretty clear counterexample to the article's claim - it doesn't matter if you're a capitalist or not, you can get great value from the technology. Funnily, the article totally avoids mentioning SD.
A1111 and the extensions system they've built is really amazing for putting FOSS into nontechnical consumers' hands.
I’m asking, I have no idea. I was under the impression that MJ was a easier to use toy. I didn’t know they were close let alone Mj superior.
I personally found that stable diffusion has been just as good at producing imagery. And i have seen plenty of fine tuning done for specific purposes (see https://civitai.com/models/5585/deliberate-for-invoke for example).
And the fact that SD is available locally, as opposed to midjourney, makes it more of a hotbed for new innovations.
No matter how you slice it, the rich gets richer, those in power centres in government gets more power, and the majority in the lower middle class loses the most.
Y = K^(1-x)*L^(x)
Here's ONE WEIRD TRICK that CAPITALISTS HATE. Increase X.
(I think, I have no idea what the copyright rules are and don’t care, not an artist, but I can make crazy good pictures now)
Barely 10 days holiday per year, no universal healthcare, car-infested cities, unsafe cities filled by homelessness, racial tension all over the country, drug addiction, and many more. The US is a living nightmare yet people like you are blinded to think you're still the best in the world. Pity.
FTFY
Even if AGI itself was “good”, that does not mean it will be used or deployed fairly and equitably. In fact, I’d posit without major regulation it will just increase the wealth gap
Russia was dirt poor in 1917. They became a super power in 30-40 years moving an enormous amount of people out of poverty and lack of modern niceties.
China changed in 78 enough already. How do Cuba or Vietnam survive? How did they survive when heavily sanctioned and attacked by capitalist westerners? We have yet to see a capitalist country do that.
I can't speak for grandparent, but when I say "communist" disparagingly, I speak as someone who was born and raised in communist Czechoslovakia, who remembers actual communism.
I remember being in kindergarten, learning how there was a great, kind man Lenin who loved children. And I was already somehow thinking, something is off: this shit shouldn't be taught to kids like us.
So you kids can crumple that up in your woke pipes and smoke it. Pardon me, vape.
People mistake oppression and basically incompetence of the soviet union as proof that communism will never work, but it wasn't communism. During the Lenin era, industrial output actually far outpaced western countries, millions were lifted out of poverty. Same in China, Mao's madness with the great leap forward plunged the country into famine. These weren't communism.
There has not been a real communist society in human history. It'll require the population to change its mindset first from greed, ignorance and blind faith, towards critical thinking, socialism (care for others). These are pre-requisites for any ideals, even democracy.
The falacy of the neoliberal order is to think greed, excess consumption and the paramount right of the individual will magically bring a prosperous society, where in fact it sets up an exploitative system where the vast majority toils for the tiny minority and stays poor and stupid.
Your argument is a typical one. See this 18 min video for why that isn't necessarily true: https://www.youtube.com/watch?v=nbkMDb1jJCw
http://beelineblogger.blogspot.com/2016/01/how-supermarket-v...
I bet this author thinks Russia isn’t where it is today because of capitalism and the US. Or Putin isn’t a result of US intervention of capitalism in Russia.
The Soviets and Socialists are anti-capitalists. How are you defending OP/OP's link that attempts to show the evils of socialism/Soviets but don't know the Soviets and socialists are anti-capitalists?
> Besides, given how expansively I've seen "capitalism" defined on the internet by its detractors, conservation of energy qualifies as capitalist oppression.
Capitalism is the means of production being owned by private individual capitalists. Think private property. There's one basic definition of capitalism. None of this was difficult to learn once I did some reading. Or even just watching some non-mainstream and non-right wing videos. Just like I've done with media from the mainstream and the right to see where they are coming from.
this wasn't the case for other capital goods that was invented in the past, so why is AI going to be different?
In the past, other forms of automation brought on more wealth, which did enrich the rest of society (on top of bringing personal wealth to the founders).
The only difference is that _your_ current view has been normalized to see the products of that capital goods as normal!
It really seems like “all of human history” tells a different story. The standard of living for virtually everyone is much higher now than it was in the past. Obviously there are people who capture a large share of the value. That is true is in any competitive system. But technological advancements have improved the entire world on average.