Currently, it's very expensive and most companies are running at a loss to grab market share. Should tax payers fund these losses as well? People like Bernie Sanders only want to benefit from a situation where all risk is taken out of the equation and they can take the wins, which doesn't line up with reality.
In addition to this, it's relatively cheap for the average person to get access to AI, since it's essentially being subsidized by investors.
Are we going to pretend that this is what's in the article, or are we going to engage with the substance of the article?
AI development is not purely a private-sector achievement, it never was. Governments fund research, infrastructure, education, energy systems, procurement and industrial policy that make large-scale AI possible. So taxpayers already absorb part of the cost and risk, while private companies seek to keep the resulting assets and profits for themselves.
If we look at how (or on what) AI is built, it becomes obvious that it's not even a public funding matter only, since these tools effectively extract humanity's accumulated language, writing, research, culture and knowledge. Not private info, not private resources, not some person's secret invention, but our shared knowledge. The companies have enclosed a collectively produced resource and converted it into private property.
Furthermore, we either accept that IP is legitimate / a strong argument, in which case humanity should have a easy claim over the knowledge used to build AI, or that IP is weak / a bit of a joke, in which case AI companies' claims to exclusive proprietary control are automatically weak and we have even more reason to nationalise these services.
No, Jacobin, a bunch of us do not welcome it, for what seems to us to be very good reason.
"Nationalize" could mean two things: Either the government pays, or the government takes.
If the government just takes, that runs afoul of the last clause of the Fifth Amendment of the US Constitution. "... nor shall private property be taken for public use, without just compensation." If the government just takes, that's dead. Since it's essentially impossible to get a Constitutional amendment to pass, that means that the Constitution is being violated. If that clause of the Bill of Rights can be ignored because Jacobin feels like it, how much of the rest of the Constitution still stands?
And if they can just take those companies, then they can just take anything else they feel like taking. Private property is dead.
I do not welcome the death of the Constitution and private property, no matter how much Jacobin thinks I should.
On the other hand, if the government pays for it, then we have a completely different set of problems. We have, first, the problem of paying for it. Where is the money going to come from? It's either going to come from deficit spending, or from taxes. And deficit spending has a way of showing up as inflation sooner or later.
The other problem with the government paying for it is paying for snake oil. If I create a business, and claim it's going to be revolutionary, that it's going to change the whole economy, that it's going to make work obsolete, and if I hype it well enough to convince the government, I may get them to buy it from me, whether or not it is all that I say it is. And the better I hype it, the more it appears to be worth, so the more they should pay for it.
In fact, at this time it is unclear whether AI companies are worth the massively high valuations that they are going for. (Of course, if they are, if the government wants to buy them later, that would take even more money.)
I don't want to reward hype artists by overpaying for their puffery, I don't want more deficit, and I don't want more taxes.
So no, I don't welcome nationalizing AI, no matter how much Jacobin tries to tell me that I should.
Now, if you want to claim that AI has taken from the public commons for training data, that's a fair argument. I'm not sure that it turns into a legally defensible claim on the companies, though, under current law. And if not, then we're left with "nationalize" meaning what it traditionally means. And I'm against it.
"the wealth it generates must benefit humanity."
Currently, it's very expensive and most companies are running at a loss to grab market share. Should tax payers fund these losses as well? People like Bernie Sanders only want to benefit from a situation where all risk is taken out of the equation and they can take the wins, which doesn't line up with reality.
In addition to this, it's relatively cheap for the average person to get access to AI, since it's essentially being subsidized by investors.
Are we going to pretend that this is what's in the article, or are we going to engage with the substance of the article?
AI development is not purely a private-sector achievement, it never was. Governments fund research, infrastructure, education, energy systems, procurement and industrial policy that make large-scale AI possible. So taxpayers already absorb part of the cost and risk, while private companies seek to keep the resulting assets and profits for themselves.
If we look at how (or on what) AI is built, it becomes obvious that it's not even a public funding matter only, since these tools effectively extract humanity's accumulated language, writing, research, culture and knowledge. Not private info, not private resources, not some person's secret invention, but our shared knowledge. The companies have enclosed a collectively produced resource and converted it into private property.
Furthermore, we either accept that IP is legitimate / a strong argument, in which case humanity should have a easy claim over the knowledge used to build AI, or that IP is weak / a bit of a joke, in which case AI companies' claims to exclusive proprietary control are automatically weak and we have even more reason to nationalise these services.
> Should tax payers fund these losses as well?
should governments make risky investments is a good question, but lets at least tax unrealized gains and wealth first.
No, Jacobin, a bunch of us do not welcome it, for what seems to us to be very good reason.
"Nationalize" could mean two things: Either the government pays, or the government takes.
If the government just takes, that runs afoul of the last clause of the Fifth Amendment of the US Constitution. "... nor shall private property be taken for public use, without just compensation." If the government just takes, that's dead. Since it's essentially impossible to get a Constitutional amendment to pass, that means that the Constitution is being violated. If that clause of the Bill of Rights can be ignored because Jacobin feels like it, how much of the rest of the Constitution still stands?
And if they can just take those companies, then they can just take anything else they feel like taking. Private property is dead.
I do not welcome the death of the Constitution and private property, no matter how much Jacobin thinks I should.
On the other hand, if the government pays for it, then we have a completely different set of problems. We have, first, the problem of paying for it. Where is the money going to come from? It's either going to come from deficit spending, or from taxes. And deficit spending has a way of showing up as inflation sooner or later.
The other problem with the government paying for it is paying for snake oil. If I create a business, and claim it's going to be revolutionary, that it's going to change the whole economy, that it's going to make work obsolete, and if I hype it well enough to convince the government, I may get them to buy it from me, whether or not it is all that I say it is. And the better I hype it, the more it appears to be worth, so the more they should pay for it.
In fact, at this time it is unclear whether AI companies are worth the massively high valuations that they are going for. (Of course, if they are, if the government wants to buy them later, that would take even more money.)
I don't want to reward hype artists by overpaying for their puffery, I don't want more deficit, and I don't want more taxes.
So no, I don't welcome nationalizing AI, no matter how much Jacobin tries to tell me that I should.
Now, if you want to claim that AI has taken from the public commons for training data, that's a fair argument. I'm not sure that it turns into a legally defensible claim on the companies, though, under current law. And if not, then we're left with "nationalize" meaning what it traditionally means. And I'm against it.