I'm struggling to see how it can be legally possible (never mind moral) to take people's money on the basis of X and then do Y. In the UK, that used to be called "obtaining pecuniary advantage by deception" [https://en.wikipedia.org/wiki/Obtaining_pecuniary_advantage_by_deception], but now it's simply called "fraud". Prior to 2019, OpenAI accepted donations (a pecuniary advantage), and hired researchers at lower than market value salaries (also a pecuniary advantage), on the basis that it was a non-profit company, but has since transitioned into a for-profit. How this is not obtaining pecuniary advantage by deception (i.e. fraud) I simply don't know. What am I missing?
OpenAI is a big Example of this bullshit phenomenon, but the subscription economy is another. Or any other reneged promise. If you "bought" a show/movie hosted by a streaming service and it was yoinked, you're out of luck. Car manufacturers are all racing to the bottom of the barrel - some will give you a free trial of "remote start" or "heated seats" and then abruptly tell you to buy a subscription to keep using these "premium features". I'm convinced Office Space was a time travel movie based solely on the printer ink eldritch abominations that emerged later.
Fraud is an apt name for it and we shouldn't allow it at any scale.
It's legally possible in California but for a company like OpenAI it will require the consent of the Attorny General. The main issue will be what to do with OpenAI assets if these are, as I assume, held in a public trust. These cannot be simply converted into the legal ownership structure of a for-profit company. This is the point in the letter Gary links to. They aren't making this up.
OpenAI started as a non-profit. All of its donors were businessmen, investors, or engineers. There was no explicit prohibition on commercialization, it just made sense back then for OpenAI to be a non-profit research lab.
Once they got some promising results, and realized that without a lot of funding they won't get anywhere, they chose to seek business investment, in return for future profits.
All fair and square. All they need to do now is to ensure they pay taxes properly given the different rules for profit-vs non-profit.
OpenAI may make that argument in court, but it's flawed, so good luck with that. Being a non-profit (BTW, I run one) doesn't mean not being able to commercialise, it means that the organisation in question is a social enterprise organised NOT for the financial benefit of shareholders, but for the benefit of a particular community (such as all of mankind) via particular means (such as the development of benevolent AGI), which means that any surplus (i.e. profits) must be retained by the organisation for the purposes of pursuing its stated mission rather than being distributed to shareholders. That was the basis on which OpenAI accepted its pre-2019 pecuniary advantages, without which it would never have achieved its initial successes (particularly GPT-2) on which the entire company was built.
This is all good. The question is, can a non-profit be allowed to change its status, with the consent of all (or at least a large majority) of stakeholders and staff?
The motivation here is, of course, that if you want to achieve great results, you must have a lot of cash to do R&D, and then it also make sense to sell your product and get investors.
I understand the motivation for reconstituting from non-profit to for-profit (although it's naive to imagine that this can be achieved without corrupting the original objective). If OpenAI genuinely secured the prior written agreement of ALL of the original non-profit stakeholders before reconstituting the company as a for-profit then I don't see how there could be any legal or moral issues. However, I don't believe that that was the case.
As I know, the only person who objected was Musk. He likely put in more money than the rest.
But even with Musk, despite what he said much later, back then, when he left OpenAI, the issue more of control and leadership, rather than non-profit.
All we can do when there are disputes, is to follow the rule of law. Otherwise there is chaos. There will always be unhappy people. OpenAI will become a for-profit by painstakingly following all applicable procedures.
20% of the speculative 150B valuation comes to 30B but that's not cash in hand, just all hot air!
Much of MS investments in the billions has been in compute credits, they probably don't have much in cash otherwise they would not be seeking financing. All they got are the "IP" (code and weights) and "talents". "Talents" can bail any moment as have been going on. So I like to see they open up their "IP". (Nothing is free they got to give up something!) But I suspect there is not much there to see either, just some cute engineering around a transformer blackbox. Maybe that's why they don't want to open up as there then goes the aura around sam's head. The point is, no one should get special treatment, not even "market leader". Play by the rule and be fair.
Since "OpenAI Nonprofit’s Valuation at North of $30 Billion", by whatever calculus that was arrived at, OpenAI must pay up an actual $30 billion of cash.
Never mind that valuation has nothing to do with revenue and profit, and is a speculative measure based on future performance. Currently OpenAI makes 3 billion in revenue, but the loss is 5 billion.
This is bad logic which would run businesses into the ground before they even have a chance.
Adam Smith's warning about — when push comes to shove — the singleminded focus of entrepreneurs on their own profit (the positive effects for society are are a side effect, just as the negatives are) rules again. You cannot drive 'values' via a single 'value' (money).
totally agree...Gary keep up the posting and pressure for AI being so much more than just profit pursuit which will be it's downfall if that is the final goal
This is another example of free market entrepreneurs wanting to receive business “welfare” which is ultimately paid for by the tax-paying public. I’d recommend no less than a 30% share for the government or a government corporation, requirement to go publicly traded, and regulated caps on C-suite salaries and bonuses. It would still be less than a VC shark would have gotten. And oh yes, all development to date to remain in the public domain. Cheers!
Not sure how this exactly works but I feel there might be much incentive for lobbying from both openai and investors to allow transformation from non-profit to for-profit without requirements of giving up assets or opening up "IP"s. After all, why would investors want to invest on the premise of for-profit designation which may lead to loss of "IP" and thus value of target of their investment? If that is what comes to pass, then I hope all future B school curriculum be enriched with such textbook
business case where rules can be bent wherever greed goes. Pupils need the ideals, they need practices more.
I'm struggling to see how it can be legally possible (never mind moral) to take people's money on the basis of X and then do Y. In the UK, that used to be called "obtaining pecuniary advantage by deception" [https://en.wikipedia.org/wiki/Obtaining_pecuniary_advantage_by_deception], but now it's simply called "fraud". Prior to 2019, OpenAI accepted donations (a pecuniary advantage), and hired researchers at lower than market value salaries (also a pecuniary advantage), on the basis that it was a non-profit company, but has since transitioned into a for-profit. How this is not obtaining pecuniary advantage by deception (i.e. fraud) I simply don't know. What am I missing?
You are struggling to understand because you still got morals. There are others who want to give fraudsters chances. We got diversity here!
OpenAI is a big Example of this bullshit phenomenon, but the subscription economy is another. Or any other reneged promise. If you "bought" a show/movie hosted by a streaming service and it was yoinked, you're out of luck. Car manufacturers are all racing to the bottom of the barrel - some will give you a free trial of "remote start" or "heated seats" and then abruptly tell you to buy a subscription to keep using these "premium features". I'm convinced Office Space was a time travel movie based solely on the printer ink eldritch abominations that emerged later.
Fraud is an apt name for it and we shouldn't allow it at any scale.
It's legally possible in California but for a company like OpenAI it will require the consent of the Attorny General. The main issue will be what to do with OpenAI assets if these are, as I assume, held in a public trust. These cannot be simply converted into the legal ownership structure of a for-profit company. This is the point in the letter Gary links to. They aren't making this up.
What I meant was, how could they do so unilaterally? Thanks for the clarification!
OpenAI started as a non-profit. All of its donors were businessmen, investors, or engineers. There was no explicit prohibition on commercialization, it just made sense back then for OpenAI to be a non-profit research lab.
Once they got some promising results, and realized that without a lot of funding they won't get anywhere, they chose to seek business investment, in return for future profits.
All fair and square. All they need to do now is to ensure they pay taxes properly given the different rules for profit-vs non-profit.
OpenAI may make that argument in court, but it's flawed, so good luck with that. Being a non-profit (BTW, I run one) doesn't mean not being able to commercialise, it means that the organisation in question is a social enterprise organised NOT for the financial benefit of shareholders, but for the benefit of a particular community (such as all of mankind) via particular means (such as the development of benevolent AGI), which means that any surplus (i.e. profits) must be retained by the organisation for the purposes of pursuing its stated mission rather than being distributed to shareholders. That was the basis on which OpenAI accepted its pre-2019 pecuniary advantages, without which it would never have achieved its initial successes (particularly GPT-2) on which the entire company was built.
This is all good. The question is, can a non-profit be allowed to change its status, with the consent of all (or at least a large majority) of stakeholders and staff?
The motivation here is, of course, that if you want to achieve great results, you must have a lot of cash to do R&D, and then it also make sense to sell your product and get investors.
I understand the motivation for reconstituting from non-profit to for-profit (although it's naive to imagine that this can be achieved without corrupting the original objective). If OpenAI genuinely secured the prior written agreement of ALL of the original non-profit stakeholders before reconstituting the company as a for-profit then I don't see how there could be any legal or moral issues. However, I don't believe that that was the case.
As I know, the only person who objected was Musk. He likely put in more money than the rest.
But even with Musk, despite what he said much later, back then, when he left OpenAI, the issue more of control and leadership, rather than non-profit.
All we can do when there are disputes, is to follow the rule of law. Otherwise there is chaos. There will always be unhappy people. OpenAI will become a for-profit by painstakingly following all applicable procedures.
20% of the speculative 150B valuation comes to 30B but that's not cash in hand, just all hot air!
Much of MS investments in the billions has been in compute credits, they probably don't have much in cash otherwise they would not be seeking financing. All they got are the "IP" (code and weights) and "talents". "Talents" can bail any moment as have been going on. So I like to see they open up their "IP". (Nothing is free they got to give up something!) But I suspect there is not much there to see either, just some cute engineering around a transformer blackbox. Maybe that's why they don't want to open up as there then goes the aura around sam's head. The point is, no one should get special treatment, not even "market leader". Play by the rule and be fair.
Absolutely something people should be discussing….
Since "OpenAI Nonprofit’s Valuation at North of $30 Billion", by whatever calculus that was arrived at, OpenAI must pay up an actual $30 billion of cash.
Never mind that valuation has nothing to do with revenue and profit, and is a speculative measure based on future performance. Currently OpenAI makes 3 billion in revenue, but the loss is 5 billion.
This is bad logic which would run businesses into the ground before they even have a chance.
Adam Smith's warning about — when push comes to shove — the singleminded focus of entrepreneurs on their own profit (the positive effects for society are are a side effect, just as the negatives are) rules again. You cannot drive 'values' via a single 'value' (money).
OpenAI is a Ponzi scam operation. https://davidhsing.substack.com/p/openai-is-a-ponzi-scam-operation
totally agree...Gary keep up the posting and pressure for AI being so much more than just profit pursuit which will be it's downfall if that is the final goal
This is another example of free market entrepreneurs wanting to receive business “welfare” which is ultimately paid for by the tax-paying public. I’d recommend no less than a 30% share for the government or a government corporation, requirement to go publicly traded, and regulated caps on C-suite salaries and bonuses. It would still be less than a VC shark would have gotten. And oh yes, all development to date to remain in the public domain. Cheers!
This was obviously the master plan all along.
It’s just that the vast majority of the principals (you know, the ones with principles, who have now left) didn’t understand the plan.
Much chance of OpenAI having $30B cash in hand any time in the next decade or so.
A good idea, however.
We also have to remember that they will not be paying any tax for years to come, given being an excellent loss-making company.
Even as a “profit” company, they will still essentially be a “non-profit”.
But maybe it’s better to be a “nonprofit profit” than a “nonprofit nonprofit”
well no shit it's better for them, but is it better for us? should we let them?
Not sure how this exactly works but I feel there might be much incentive for lobbying from both openai and investors to allow transformation from non-profit to for-profit without requirements of giving up assets or opening up "IP"s. After all, why would investors want to invest on the premise of for-profit designation which may lead to loss of "IP" and thus value of target of their investment? If that is what comes to pass, then I hope all future B school curriculum be enriched with such textbook
business case where rules can be bent wherever greed goes. Pupils need the ideals, they need practices more.
Probably not much chance of a claw back if the appropriate people have "incentivized" to resist this solution.