Our structure
openai.com
openai.com
Fifth, the board determines when we've attained AGI. Again, by AGI we mean a highly autonomous system that outperforms humans at most economically valuable work. Such a system is excluded from IP licenses and other commercial terms with Microsoft, which only apply to pre-AGI technology.
Whenever AGI happens — and assuming this isn't a singularity-type change, and assuming that the AI is friendly rather than a paperclip optimiser etc. — the meaning of real power and wealth will change again, even though money (as currently understood) and land will still exist.
Russia is the biggest country in the world, but at ~1.7% of global GDP (and ~1/16th of the USA's) it's not even close to being a world-leading economic powerhouse. Post-AGI, I expect that even the company with the highest earnings or market capitalization will be to the dominant {entity related to AGI but not necessarily the AGI itself because it doesn't have to be a person or an agent in its own right for this argument} as Russia is to the US.
YMMV
1) Abandoned cities are extremely rare throughout all of history. Most medium to large cities outlive the countries in which they were founded.
2) Even if this were the case, it doesn’t actually disprove the claim. All that’s evidence of is that occasionally, some forces can overcome positive pressure on prices. Yes systemic failure can sometimes (very rarely) knock a city out of existence, but this is very obviously extremely rare. Even Detroit is dramatically more expensive than it has ever been. The price of an acre in Wyoming is higher than it’s ever been.
3) That’s still assuming that abandonment is a definite sign of falling land prices, which isn’t true at all. Vacancy results from the inability to produce more value on the land than the land costs. Sometimes this is because the prices rise and sometimes because the cost of production rises, but never because the price of land went down: that would just mean more profit which means more reason to stay.
you can say "oh this statement is not supported" with rebuttal, but the rebuttal has "stock market goes up" built in.. this is not true over long periods of time generally due to war or other disaster.
more profit - more reason to stay.. Somehow we are not acknowledging Walmart here. In the USA, small towns functioned on having stores and other services.. sixty years later.. whole towns are dying due to a nearby Walmart, where people drive an hour+ to shop.
neither my statement or the rebuttal has all the relevant parts.. under construction. dynamic systems
AI, like the drill press or cotton gin, will increase how much value you can produce per square foot.
Landlords will reap the bulk of the benefit, we will still have people in completely destitute poverty and the bulk of people living roughly paycheck to paycheck.
It may also go up in relative terms, but yes it will go up in absolute terms too.
Very self evident: how is it that people paid for a place to live 1,000 years ago, and now there’s been all this technological productivity gains (several billion-fold?), and yet people still spend 30%+ of their productive output on a place to live?
Been thinking the same recently, but those specific numbers are a mix of wrong and misleading.
For wrong: productivity per person has only gone up by a factor of about 50x-100x, not billions.
For misleading: the homes were terrible back then. A thousand years ago (Beowulf-era Old English), a "hall" was simply a large room enclosed by a roof and walls. In Anglo-Saxon England (the latter end of which was still 1000 years ago), simple single-room buildings, with a lone hearth in the middle of the floor for both cooking and warmth, were the usual homes of a lord of the manor and his retainers, with the entire community eating and sleeping in that single room. Even just keeping those places warm in winter was a major effort. (I've seen a few of these at different stages of development at https://www.wealddown.co.uk, which just happened to be close to where I grew up as a kid; you may recognise some of the pictures from the first season of Good Omens).
The fancy castles and complex manor homes we like to think of from Ye Olde Tymes weren't normal homes for individuals any more than the White House is today, and almost all the buildings that survive were (in addition to being lucky) important enough to be built upon over the thousand years between then and now.
Two thousand years ago, by contrast, you could get a Roman Villa in Ye Olde England that had working toilets, running water, sub floor heating, mosaic decoration tile work, and more .. all for little more than the risk of an occasional onslaught by the woman next door and her bloody scythed chariot.
""" It is difficult to give even rough comparative values for money from before the 20th century, as the range of products and services available for purchase was so different. During the republic (509 BC–27 BC), a legionary earned 112.5 denarii per year (0.3 denarii per day). Under Julius Caesar, this was doubled to 225 denarii/yr, with soldiers having to pay for their own food and arms,[15] while in the reign of Augustus a Centurion received at least 3,750 denarii per year, and for the highest rank, 15,000 denarii.[16] """
Which corresponds to 13 years 4 months gross pay in the best case of the highest earner and cheapest villa, or 17,777 years for the worst case combination from those numbers.
[0] https://history.answers.com/ancient-history/How_much_did_hou...
[1] https://www.unrv.com/culture/roman-houses.php
[15] Adkins, Lesley; Adkins, Roy A.; Adkins, Both Professional Archaeologists Roy A. (2014-05-14). Handbook to Life in Ancient Rome. Infobase Publishing. ISBN 978-0-8160-7482-2. Archived from the original on 2023-09-04. Retrieved 2022-04-14. - https://web.archive.org/web/20230904145101/https://books.goo...
[16] appears to be the same link as 15, don't know why it's listed separately on Wikipedia
I suspect an upmarket villa rarely ended up in the hands of many who started out in the lowest ranks (save for a talented|lucky few grifters with good side earners going on while enlisted), connected sons from well off families who went straight to mid ranking | aide de officer roles and picked up "gifted lands" for service etc were perhaps more likely to get a swanky house on an estate.
That aside, the various Time Team and British digs that uncovered villas in the UK on estates found some very fancy, almost modern, housing with spas, plunge pools, stone ducted heated, etc.
The later Danes seem to have dropped the bar on quality of housing.
real: late Middle English (as a legal term meaning ‘relating to things, especially real property’): from Anglo-Norman French, from late Latin realis, from Latin res ‘thing’.
An "estate" writ large is many things, including social and economic relationships, contracts, and instruments that are more ideas and behaviors than an actual "thing". Real in the sense of "real estate" means the physical part of an estate; the "things" you can touch. The real in real estate is not denoting what is the actual, *realized* value of the estate, which I think is what you're intimating.
Another interpretation is that if AGI is achieved it will be open available beyond OpenAI and accessed through other sources.
What a sad testament to the current zeitgeist that economically viable is our bar for AGI.
Van Gogh was not “economically valuable” for the span of his life.
"most" but not all economically valuable work.
ChatGPT (or what follows) can't swing a hammer and it can't apply spackle. Maybe I should start looking into the trades.
If it doesn't produce or dramatically aid in the production of Food, Water, Shelter, Energy, or Transport... it will never be as intrinsically valuable. The vast majority of work in those categories requires irreplaceable physical labor.
We are probably not going to write it, but I assume as long as we are in control, we are going to drive the design. Software is most certainly the end game.
1. Apple 2. Microsoft 3. Saudi Aramco 4. Alphabet (Google) 5. Amazon 6. NVIDIA 7. Meta (Facebook) 8. Berkshire Hathaway 9. Tesla 10. Eli Lilly 11. TSMC 12. Visa
How many would you reckon are dramatically involved in "Food, Water, Shelter, Energy, or Transport" in a way that software can't enhance, multiply or replace?
Apple, Microsoft, Alphabet, Amazon, NVIDIA, Meta, Berkshire Hathaway, TSMC, and Visa do not produce anywhere near the same kind of core, intrinsic value. We have a fundamental disconnect between monetary value and actual value to society. I'm suggesting that our capital has been invested extremely poorly for at least the last two decades - chasing easy software wins instead of building hardware we need to solve real problems and putting humans to work to help construct it. These companies build products and services that enable business, commerce, hardware and software that enhance our experience in the world as we know it (including those core categories), but the world as we know it could change in an instant.
If everything were to go to shit, the most valuable are Food, Water, Shelter, Energy, or Transport (and as another poster pointed out, probably Art although it's kind of in a league of its own). They've always had the most intrinsic value. And while we need software to enhance a lot of these, the physical laborers are far less replaceable and therefore far more valuable. Technology is much farther away from replacing people in meatspace than cyberspace.
Apple is not stupid - they are probably best positioned to dominate computing over the next decade at least. They are computing company that specializes in enhancing experience, but there is a looming energy crisis and they know it. That's why their CPUs are focusing so heavily on performance per watt. We are going to need radically more efficient hardware that is cheaper to produce to meaningfully deploy computers to solve out problems in the future.
If datacenters can not become sourced entirely by renewables, the your pocket computer (and personal AI) is going to be more sustainable and therefore more valuable.
Edit: heck I believe that if you have a car moving at 1cm per 20 second that's sending a picture every 20 second to gpt4v and some glue code to send and receive from gpt4v it'll will be able to self drive.
Agreed. My original comment was simply that ChatGPT (or whatever) can't swing a hammer yet.
The whole thing feels like an oncoming tsunami and I'm just trying to figure out how to scramble to higher ground (while it lasts).
They can even do multi-modal transfer learning: have deviations from a previously learned process explained in words it's diagrams, then start directly zero-shot-like at a performance level that won't waste ingredients, with all the normal learning speed aspects still remaining.
Only thing that could be bad if future GPT releases are withheld from Microsoft. This is a worst case, but possible scenario, according to the structure of OpenAI. Microsoft might have an agreement for a length of time or profit to still get GPT 5 but maybe not GPT 6.
At that point, Microsoft is probably working on their own systems, and have enough time to make a competitor. Also, they can fund competition and get more access to models and stay ahead.
Overall situation for Microsoft is just doubling down on their innovation and investments, which is what Satya has already done and will keep doing, and is what has gotten Microsoft to the 3T market cap.
A comparatively massive user base, an app deployed onto many phones, accounts, the cutting edge AI models, the data, contracts for exclusive data access, compute contracts worth billions at a time when GPU compute is getting seriously scarce, corporate GTM partnerships from Bain to morgan stanley to Duolingo, a powerful brand (ChatGPT), tens of thousands of apps already built on their APIs, benefits from Government Lobbying
Ya, so really, what do they have, it sounds like they are basically dead.
Microsoft signed a contract that reads like a Kickstarter project ("would be wise to view any investment as a donation") because of what OpenAI had.
ChatGPT is a brand that has almost become as powerful as Google is for Alphabet and what Windows is for Microsoft and the 'i' prefix for Apple.
It's gonna be really hard to get people as excited about AI as ChatGPT because you need a similar combo of technical leap + penetration in pop culture, in order to get everybody.
Basically the same concept that is not well understood by that dude who changed twitter name because he bought a one letter domain back in the 1980s
On a literal reading, this definition of AGI seems impossible to meet without extensive breakthroughs in robotics. Even today, most work involves a physical component. Those of us who wrote text for a living are a small minority.
AGI would solve that problem.
What you’re saying is, the bot needs to be 1 million times better at writing software than John Carmack to reach OpenAI’s definition of AGI, because it needs to be able to produce software that no one else has even come close to producing. That’s a much higher bar.
https://in.ign.com/playstation-4-neo-ps4/148831/news/a-fully...
With more universal machines (powered by AGI), with human-like hands and maybe legs for outdoors, I think cost will come down due to economics of scale and the ability to repurpose machines easily.
I wonder if this could be at the crux of the fissure; does Ilya see AGI coming soon, and want to pull the rug on Microsoft? Maybe not immediately but perhaps he was flagging it. (Or maybe he thinks GPT-5 clears the bar, which would be quite the claim.)
That feels like a hint at something larger than what we've seen publicly announced so far. It doesn't seem to me like anyone should be discounting this theory.
If you want to say what you think is important about an article, that's fine, but do it by adding a comment to the thread. Then your view will be on a level playing field with everyone else's: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...
(Submitted title was "Microsoft is a minority owner of OpenAI Global LLC")
Third box down says "Holding company for OpenAI non-profit + employees + investors".
What do they mean "for OpenAI non-profit". Is there another non-profit? Or when they say "for", they mean the holding company is "for the benefit of" the non-profit?
I first read it as "this is a holding company for the OpenAI non-profit", but that makes no sense because the OpenAI non-profit (above it) already owns the holding company.
The diagram is intended to be self-explanatory and follows the convention of normal organizational depictions.
>What do they mean "for OpenAI non-profit". Is there another non-profit? Or when they say "for", they mean the holding company is "for the benefit of" the non-profit?
Yes, the holding company is for the benefit of the non-profit AND employees AND investors. The holding company "investors" would cover all the pre-Microsoft early funders, eg Sam, Musk, Thiel, Hoffman, Bezos, ect.
>but that makes no sense because the OpenAI non-profit (above it) already owns the holding company.
The distinction is that the non-profit is not the only owner of the holding company. Employees and early investors have no ownership of the "OpenAI 501(c)3 non-profit", but they do have ownership of the holding company.
The holding company then shares ownership of the for-profit "OpenAI global" with Microsoft.
If bad actors abuse the tech, then you make laws (or, in many cases, already have them on the books) against those behaviors.
The negative externalities of AI are called "crimes" and we have a framework for handling them.
Somebody might be thinking they're gods. At any point however government can nail them
Edit: I wonder if OpenAI meta-advises itself with ChatGPT or how their attitudes towards that notion are or evolve over time. I feel like there always needs to be a head to fly should disaster or inconvenience strike the C-Suite...
Edit: where would the line between conflict of interest and proving itself/concept fall out of curiosity. Is there any way for that to comport with their fiduciary responsibillities even if it is provably superior at the PKIs (if CEOS even have that outside of profit/loss etc)
A non-profit parent entity with a wholly owned for-profit subsidiary (actually, the non-profit wholly owns and controls a third "manager" entity which in turn is the controlling but not majority shareholder of the for-profit subsidiary) where the profit of the for-profit entity is "capped" and oh by the way, that for-profit subsidiary has a 49% shareholder who happens to be Microsoft is a Rube Goldberg contraption.
The above is just scratching the surface of the absurdity that is the OpenAI corporate structure. Microsoft despite having 49% ownership of the for-profit subsidiary gets 75% of their profit..........until their initial investment is recouped. Employees get Profit Participation Units (PPUs) which are NOT equity but which entitle employees to a percentage of profits, again up to the "cap". I have a hunch that when they say Sam and the Board have no "equity" they mean that they have no traditional equity in the form of RSUs or options, but instead have a similar PPU arrangement. Then of course, when OpenAI achieves AGI, whatever the hell that is is, then everything from that point forward reverts back to the non-profit entity and Microsoft, other investors, and employees get hosed.
They planted the seeds of their own destruction by ousting the CEO and causing the Chairman to quit shortly thereafter. Lawyers for Microsoft and other investors will be knocking on the boards' door Monday morning and some of these board members who have never had a real job in their lives will get steamrolled.
Legally, Microsoft clearly knew they were stepping into some whacky corporate structure.
If their profits are capped and they have no rights, that's what it is.
You think they're just going to complain their way out of it in court?
I'm not convinced.
When does controversial corporate stuff not result in some kind of legal action?
It doesn’t actually matter what the situation is, or if it makes sense… you just take your pack of lawyers and you point them at the enemy and tell them to do their jobs.
I mean, have you met Microsoft?
Yeah. They have a lot of lawyers.
I think you can take it as a given this will have some kind of response… it’ll just be a matter of them trying to judge the benefits vs. the cost of it.
Ie. it is not that they can’t lodge legal action, because, trust me, they can… but they, I guess, will wait and see what Sam does, and try to decide if/who they want to back before they do anything.
Truth. And further leverage depending on how much of the $10B in cash or Azure equivalents MSFT has already delivered.
I’ve got some experience in not for profit boards and for profit boards. For profit boards serve shareholders. Not for profit boards are often the ultimate authority. There are no shareholders. They may have members who fill that role, or they may not. They can also be essentially self-sustaining, with only board members electing new board members.
Unless given some authority by contract Microsoft likely has no ability to do anything with their board.
Also, "hiring the best lawyers in the world" is similar to the "nine women can't give birth to one baby in one month", that is, from a certain point on both sides can "hire the best lawyers in the world", or pretty much close to it, an order of magnitude (or two or three, like MS has over whoever will stand behind the OpenAI's board members in helping them with money) bigger scale doesn't necessarily guarantee the expected outcome.
If one company could sue another smaller company out of business at will, there would only be one company.
It does happen that companies go bankrupt fighting invalid lawsuits, no doubt, but it’s rare and it’s almost never a company that just built the most talked about product of the decade and raised an 11 digit war chest.
It will be hilarious if they sold all their IP to Google or Amazon.
But it's more likely in such a case, they will just open source / toss into public domain all their IP, shut their doors, and start a new non-profit which MS has no good reason to sue.
Or that if Microsoft did sue (perhaps because something that happens in the near future violates their contracts) OpenAi just spends like 1% of their cash on legal fees. $100 million in legal fees is not an existential risk to companies who recently raised $10b.
That just isn’t how the legal system works. You can sue a small business into bankruptcy because a few million in legal fees is enough.
Depends on the contract. If the contract says nothing about OpenAI having to work on LLMs ... then there is nothing MS can do if they stopped all LLM work and worked on something else.
I think that's basically literally their charter?
Their stock is probably up $250B because of this investment...
The idea they're getting screwed is interesting.
Microsoft is a lot of things - stupid and incompetent are neither of those things.
They want to have their cake and eat it, too.
I am pretty sure this structure was not a mistake.
A non profit with income producing subsidiaries is a well established structure used by plenty of organisations globally.
This work needs a ethical steward, not a mega payout for those who well postponed to make some fast money.
Seems we’ve just seen a hard turn attempting to avoid an otherwise inevitable profit driven enshitification of Open AI’s work.
I imagine there are two factions in OpenAI, one group (let’s call them the Ilya faction) are only motivated by AGI, while the other group (maybe Sam faction?) is interested in reaping the short term profits. So why is it wrong for Ilya to complain if he thinks Sam only wants short term profits and is hoodwinking the rest of the folks into believing he is also interested in AGI? Short term profit is not what he signed up for.
A C-Corp with multiple share classes where the CEO is also a Chair with multiple seats will be superior in that affect
Neither structure really removes the profit incentives, but for people that believe that you can literally say anything to those people. You can say “our bylaws say” or “we made a B Corp because we like to be socially driven” or “we’re a non profit!”, these are emotional tugs than any legal reality
Just because MS sues, doesn’t mean they will win. If their contract with OpenAI say they don’t have control over the for-profit part of OpenAI (and they definitely have no control over non-profit part) then they don’t. The courts will rule against them and it will be just a waste of everyone’s time.
and I wouldn't be surprised if the under the table answer would be that they trusted under the table verbal promises of a certain ex-CEO and serious FOMO ;-)
So no, Microsoft Lawyer's can knock all they want.
IMPORTANT * Investing in OpenAI Global, LLC is a high-risk investment * * Investors could lose their capital contribution and not see any return * * It would be wise to view any investment in OpenAI Global, LLC in the spirit of adonation, with the understanding that it may be difficult to know what role money will play in a post-AGI world * The Company exists to advance OpenAI, Inc.'s mission of ensuring that safe artificial general intelligence is developed and benefits all of humanity. The Company's duty to this mission and the principles advanced in the OpenAI, Inc. Charter take precedence over any obligation to generate a profit. The Company may never make a profit, and the Company is under no obligation to do so. The Company is free to re-invest any or all of the Company's cash flow into research and development activities and/or related expenses without any obligation to the Members. See Section 6.4 for additional details.
I'm not sure why you think this part is absurd. This is basically a preferred return, and is very common in private equity and real estate investment.