OpenAI board reappoints Altman and adds three other directors
reuters.com
reuters.com
(I've merged those threads hither in the interest of having a hopefully more neutral article as the top link.)
Something like announced with this that their new model "Rubicon" had, under the resumed leadership of Altman, crossed a milestone threshold.
https://www.newyorker.com/news/letter-from-silicon-valley/th...
https://www.ycombinator.com/blog/meet-the-people-taking-over...
Look's like its almost 10 years since he officially joined full-time. We should have a party.
To use the parlance of this community of people: people's priors here all seem to be that something shady was going on with Altman, but the strat the board took here was so obviously doomed that they should probably be looking for hypos about the board being shady.
I am fine with people being greedy, but don't say it isn't greed.
“intense and selfish desire for something, especially wealth, power, or food.” (Oxford via Google)
Who doesn’t feel “intense and selfish desire” for money? Who doesn’t feel that going into a performance review?
It’s human nature to feel this feeling, if that’s what we want to call “greed”.
Yes, greed is human nature, just like lust etc. We value when people suppress such feelings, but we know everyone has them.
What do you think has kept life procreating on this planet in the face of overwhelming hardship? Lust.
I think it’s one-eye-blind to demonize greed, lust, or any other part of human nature. Those drives are inseparable from us being here in our current state, living reasonably well, conversing over the internet on phones and computers.
A great deal of what has driven civilization forward are the people who are driven by curiosity. People like Archimedes, Plato, Newton, Einstein. Not to mention the millions of curious nameless folks who improved the boat, the spear, the wheel, the book. Folks like RMS without whom I doubt we'd have anything like the current open source ecosystem including Linux. These tend to be people who are not only curious, but whose sense of greed is lower than the average.
Without these folks, it's not clear to me that humanity ratchets forward in progress. If you just have greed and lust, the steady state would be something like power politics of states warring against each other with swords for thousands of years. History has gotten stuck in those states many times, and when we finally break out it's typically the curiosity-driven folks who give the greed-driven folks a reason to invest their money in something other than raising an army for yet another war of honor.
I wasn’t saying greed is the only driver.
Edit: I’d also add, I think it’s a false romanticism to think that people like Einstein and other great thinkers and creators were not without the lower drives like greed and lust. If you think about all of the stages of their lives, and all of the actions and endeavors that led to them advancing their fields, greed and lust would have been interleaved as drivers of many of those actions. Maybe one year Einstein just had a bigger fire lit under his feet because he was trying to impress a girl. Or he worked harder one year because he needed the money.
Then you have people like Edison who were certainly motivated by greed and the lust for power. You have the industrialists like Carnegie who built the railroads. Then you have all of the workers across generations who did all of the work… how much of that work was driven by the “selfish and intense” desire for money?
With respect to lust, I didn't mention it because it was orthogonal, but my understanding is that they're about evenly divided into groups with higher than average lust and lower than average. Paul Erdos is a famous name rumored to be in the latter category.
I think the part you're missing is all the brilliant ideas from brilliant people we missed out on over that 2,000+ year from people we never heard of due to science and art being funded only haphazardly by random wealthy patrons.
For almost all of history it was simply not an option to make a living simply by writing or researching. That's still the case in much of the world.
But once we started to systematize it properly, and governments recognized that technology was a strategic necessity, bam, within a century or two we went from playing with primitive batteries, to nuclear fission, the internet, and now AI.
Somebody was paying for Stallman to sit in an MIT lab and write a free compiler. Maybe we don't talk enough about who or why, or how to replicate that kind of success.
if you looked at great inventors say like Nobel, who amassed a fortune based on research and innovation - would you characterize him driven by greed ?
Why ? Is he any different from say Philo Farnsworth, or Louis Pasteur who never became wealthy ? Were they not driven by the same drive as Nobel ?
Is the drive to succeed different for these sets of people?
Truth is in the eye of the beholder
People are very bad at reasoning about motives of others. Faulty syllogism goes: Nobel invented dynamite. He became wealthy. Therefore, he must have been driven by desire for wealth. Same goes for Pasteur, of course. Maybe he was driven by desire for wealth, but couldn't succeed. My Bayesian senses tell me that Pasteur not being driven by greed is more likely, but I digress.
Leaving individuals aside, I believe that a system that doesn't reward greed is a better system. When a system rewards greed, anti-social behavior becomes prevalent, because it's easier to eat others' pie rather than making a bigger cake. In a system that doesn't reward greed, I believe that curious people will make the cake bigger for the sake of it.
I think -but i'm not certain!- that we have english words for those that "eat others' pie rather" vs "making a bigger cake". I think this is greed, and driven, respectively.
But as you say, we are extremely bad at ascertaining motives of others. So either we get lazy and reduce others to bring them to our level, or we are naturally suspicious and do so anyway.
I'd love to know more of a system that doesn't reward greed. We have laws in place to stop outright theft, but at least in countries like USA and Switzerland, laws do an extremely poor job at making certain politicians that come into politics dirt-poor, stay poor, while in office and even afterwards. That to me is exactly one aspect of the system that must change.
Compare, for example, the engineering-focused Roman empire with the idea-focused Greeks.
> governments recognized that technology was a strategic necessity, bam, within a century or two we went from playing with primitive batteries, to nuclear fission, the internet, and now AI.
Governments have recognized this since at least the time of Archimedes, though. A lot of well-known mathematicians worked on things like canons, mines, railroad networks etc. So I think something else is needed to explain the explosion you're talking about other than governments recognizing the strategic importance of technology. Probably computers (and networks of computers) together with the figuring out of quantum mechanics played major roles.
Restraint and being satisfied with what you already have are the keys to stability.
Life is not overwhelming hardship my friend. It is bliss. At the end is the tiger eating your insides saying what bliss is your life, please become part of me. With words we deceive ourselves, convinced that we are separate from the living world we came out of.
But I don't work at AI or earn that kind of money, so who knows.
I mean, objectively that's what happens to 95% of people in that situation. Hedonic adaptation is a bitch.
but you're not, presumably. And when you _do_ have that sort of salary, your circle of friends and priorities will also change along with it. At least, for the average person who also makes your claim.
There's a very small positive reward from seeing the number go up, certainly not intense.
* Senior iOS, been doing iOS since the first iPod with retina display, Berlin, details on my website
A non profit is subject to labor markets just as well. They can't just go back on their implicit payment deals, in this case the payment of equity, and expect that nobody will complain.
Those employees worked under terms of employment that included the expectation that the equity would be worth something.
If openAI didn't like that, then they should have paid them in cash, without the equity.
Its not plundering to use your own labor and market power to collectively bargain for yourself.
Don't get me wrong, I'm not a Sam stan or anything.
I think you're forgetting what the entire mission of OpenAI is
Look, the board could have held the line, and then died when every one of their employees jumped ship. But nobody can say OpenAI's labor owed anything to the mission. They're all free agents, just like us. Real easy to say that someone else should take a pay cut to serve a mission you care about. It's good to want things!
Sam, Ilya, Mira, Elon, and the rest of the founding team do.
> Look, the board could have held the line, and then died when every one of their employees jumped ship.
Wouldn't contribute towards the mission, again.
> But nobody can say OpenAI's labor owed anything to the mission.
Sure, they could leave, they're not slaves. But that's the entire problem here, and why the board attempted to fire him. He's subverting the mission by putting in the exact same $ mercenary incentives that cause the whole problem OpenAI's charter is trying to prevent.
OpenAI is founded and led by a charter that profit-focused AGI arms race will lead to dangerous outcomes for all of society. If you wanted to be greedy and max your $ just go work for Meta. Don't be surprised when it's whole axis of control tries to do things that aren't "make my money go up."
Anyhow, the goal of the foundation is AGI, earning the money you need to survive until you manage that does not sound evil.
One conclusion I came to was that I didn't respect those people's ability to discern or conclude.
Did you get the argument or not, and did it make sense to you in the situation? That's all that matters.
You know how Hollywood gets computers, coding, and hacking wrong? That's how bad non-lawyers are with law.
Or, pertinently to this case — who knows more law: one of us, or ChatGPT?
We absolutely do make stuff up, we just don't realise we're doing it.
You can hire new engineers & they’d have proven they’re working to benefit humanity.
Now it’s owned by Microsoft.
Before the industrial revolution, land was paramount; after, capital. AGI is likely to be as significant a change, and make money (as we understand it today) less important in a similar way.
Even if AGI becomes some form of new capital, it is still the same as capital as we know it today. Until the day humanity transcends and we no longer face the human condition of course.
"Important", but not "just as important". Land used to be synonymous with wealth, because you needed land to support the peasants who grew the surplus that fed your army. Now? Now we have enough spare land to have national parks and substantial re-wilding programs.
Labour is still important, but also not "just as important" — all the improvements from each of the three industrial revolutions to date have been about getting more stuff done per person, and we're starting to get "lights out" factories[0].
My comparison is to say money will be "relatively less", rather than "not" — to the extent that money is a fungible token of power, sure, that's probably still going to be useful to someone at some point — but if we've all got our own personal yachts[1] with all the on-board hydroponics, bioreactors[2], and CNC manufactories needed to meet our material needs, even if money is still around, it becomes no more important to our personal experience of life than it is for people today to be personally concerned that the harvest is brought in on time or that the ore veins in the mine might run out[3].
[0] https://en.wikipedia.org/wiki/Lights_out_(manufacturing)
[1] Snow-Crash style: https://en.wikipedia.org/wiki/USS_Enterprise_(CVN-65)
That said, given we've already got some lights-out factories making more robots, it's a reasonable assumption they'd actually look more like this even without human-level AGI: https://scifi.stackexchange.com/questions/265026/why-dont-th...
[2] https://www.youtube.com/shorts/aMo2jVKmeoY
[3] Which isn't to say that nobody will worry, just that most won't. Me 15 years ago: https://en.wikipedia.org/wiki/Peak_oil
If humans really do become completely redundant, then maybe we've reached the end state of humanity and might as well just die out.
We're fed to the point we have obesity, but most of us still stop eating at a certain point.
We're oversupplied with material goods to the point that storage lockers for excess stuff is a big industry, but most of us don't hoard to that degree, and it's weird when we see people who do.
We're oversupplied with light to the point that we've got to remind ourselves to turn them down at night, but we do turn them off.
We do reach saturation points with these things, despite over-consumption.
A good way to think of it is as the embodiment of its creator. That's OpenAI, and because Sam and the whole staff have publicly acted in what we'll call a "controversial" manner that is in serious tension with the common good of all humanity, folks rightly should view OpenAI's AI black box untrustworthy, a snake oil.
Maybe the AI will habitually pit you against your trusted colleagues when doing so might be profitable to OpenAI staff, the way Sam did. Maybe the AI will claim to be altruistic and then once it's achieved a position of trust it will, when you are most vulnerable, function decisively against users' needs and goals the way OpenAI staff acted against their managers in the OpenAI board.
These people are selling trust when they have decisively demonstrated they are unworthy of it. There are entire professions built on trust that demonstrate this is not a "natural" course, it is instead a path to long-term irrelevance.
Is it possible the interests of the employees align more with the interests of Altman and that they diverged from the rest of the Board (which the Board did not expect)?
He’s a next level psychopath
I think there were no strategy, they burnt a fuse instead.
why do you presume that there must be a good party here?
- They had a huge funding (>80B?) round planned in which employees could cash-in. Without Sam that round was guaranteed to not happen. So they all had huge financial incentive to support them. Which I guess, characterizes him as a great manager who thinks many steps ahead.
All incentives all different stakeholders (employees, leadership, board, investors) have are all aligned with the full for-profit side of the organization. In fact, that’s the only side.
“Show me the incentive and I’ll show you the outcome”. Charlie Munger
"Show me the loot, and I'll show you the theft" - Willie Sutton (when asked why he robbed banks, what he actually said was "Because that's where the money is")
Immediately reminded me of Succession, where [SPOILER]: Ken initiates a proxy battle for his dad's company, then rejoins the company and attempts to "diffuse the bomb he planted".
Hard to believe this, esp. coming from board members who had experience. Even if they were right, the sudden firing was utterly unprofessional.
Yep, talk about an absolute power flex. You don't have to like it, but this is what peak performance looks like.
Note that in the US, private foundations are not permitted to own a controlling interest in a for-profit company, so the example has to be a public charity. IKEA is controlled by a Dutch nonprofit foundation, but it has no obligation to the public except to give away a small portion of the profits.
https://en.wikipedia.org/wiki/Stichting_INGKA_Foundation
https://www.economist.com/business/2006/05/11/flat-pack-acco...
The only slight downside is his position is evidently dependent on Microsoft's patronage, but I'd rather have one seat in seven and an alternative offer and the ability to threaten to take most of the company with me for the alternative offer than one seat in six with a board that believed they could oust me if they didn't like my decisions.
Certainly, no one with any sense will be “donating” any money in OpenAI without some strong and specific guarantees of what they’re going to be getting in return. So it can in no way can be considered a non-profit any more.
And if the situation is allowed to stand then I think future donors to any non-profits might think more carefully about what their money will be used for.
And what about the taxpayer? What are these guys/gals fighting over if it isn't future profits?
Has anyone else read the reporting on the non-profit's corp structure?
> OpenAI also created a subsidiary of For-Profit-LP, OpenAI LLC (let’s call that “For-Profit-LLC”), of which Microsoft is a minority owner and For-Profit-LP is the majority owner. For-Profit-LLC’s operating agreement cautions that “it would be wise to view an investment in [For-Profit-LLC] in the spirit of a donation, with the understanding that it may be difficult to know what role money will play in a post-AGI [artificial general intelligence] world.”[0]
I really have no idea why companies like Microsoft are putting so much money into a for profit entity, and calling it like a donation. Am I just skeptical or is everyone winking at each other, and acting too cute by half?
[0]: https://clsbluesky.law.columbia.edu/2024/03/05/the-untold-no...
All those "donators" went up in arms and threatened to sue/quit when they thought that those "donations" wouldn't return profits. It is a for-profit in all but name today.
They actually do things compared to this "open" and "non profit" front of a technology company.
It would be weird to even ask "How much open source code and research does the Catholic Church publish?"
The church does stuff away for free, sure, but then so too does ChatGPT have this shockingly powerful free tier.
The Catholic church has not only engaged in centuries of religious scholarship, it has given the knowledge freely to anyone. They even set up buildings where you can go and they will teach you what they have found, most people go on Sunday.
That isn't to mention their patronizing of the arts supported the greatest painters and composers that humanity has ever produced.
Now how many hungry has OpenAI fed?
https://en.wikipedia.org/wiki/Index_Librorum_Prohibitorum
https://en.wikipedia.org/wiki/Vatican_Apostolic_Archive
Also, one of the biggest issues that Martin Luther had with Catholicism was that it put the clergy between the faithful and god.
*The Thirty Years' War enters, stage left*.
> Now how many hungry has OpenAI fed?
Around as many as the number of Python programs that the Catholic Church has written with Hindi comments, which was one of the first things I tried on ChatGPT.
There's also a reason why I said "apples (openness) to oranges (doing stuff)".
OpenAI also didn't send nine crusades to liberate the holy lands from the Muslims. Someone probably thinks that makes them "woke" or something.
…as OpenAI is only a few years old, wasn't the first few years of Christianity, before Catholicism split from other denominations, "oh no, we need to stop our converts committing suicide immediately because they want to get out of their awful lives and into the heaven we promised them and they're afraid they might sin again between baptism and death if they don't off themselves"?
Even with the weird end of the "Rapture of the Nerds" (the attitude not the book, I've not read the book), I don't think we're seeing so much of that resulting from AI development. And that's despite more users than there were Europeans in 1500 AD, let alone 1 AD.
The pope is not a model.
And the pope is quite bound by ancient texts and the the pronouncements of previous popes. While no way exists to depose him if he disregards them, having the cardinals declare the current pope was illegitimately appointed and appointing a new one was fairly common back in the days.
this is actually already why non profits have difficulty raising money
many people view charities as unaccountable scams
most non profits are funded solely by the board and the endowment is managed to grow tax free. If you start reading the IRS forms 990 you’ll see pretty wild and inspirational things
While the membership is determined by the executives appointed by the board, there's at least some thought given to the idea of holding the people on the top accountable.
there is a lot wrong with that assumption based on a single anecdote.
private foundations can have a single board member and that board member can be the single donor. although the IRS does not care in granting tax-exempt status, not all states offer a single board member for their own non-profit entity catalogue, but relevant states do like Delaware. of course Delaware does right? additionally, it doesn't matter what a state's non-profit entity selection has as any entity can obtain non-profit tax exempt status from the IRS technically, even unincorporated entities like sole-proprietorships ie. your general person when operating the charitable mission.
there is some transparency of this which is not aggregated, in the 990 series of tax disclosures that non-profits have to provide, this can show how many board members or directors there are, if that's whats important for you to prove. I don't think that's important, but this is about "common" or not and I don't have the aggregate data, that's where to get partial data.
none of these forms tell you the bylaws of the organization, these are still private documents. the bylaws dictate board member selection.
next, if Donor Advised Funds count as non-profits in your mind, that's a slam dunk as those are all pools of capital for grant making run by a single donor.
and where the transparency totally fails is in exempt non-profits, these report only the "postcard" 990-N and is entirely on an honor system that they have less than $50,000 in "annual receipts". these are cheap to create and nobody knows anything about them, what's actually on their balance sheet, there are basically no audits as they are tax exempt and there is no reason to determine what's missing.
I usually don't write this stuff as its more beneficial if the people had more disjointed opinions about tax matters, but I don't think Congress will ever reach consensus on any opinion people harbor so its fine. One of those examples where truth is less familiar than fiction.
I think this is the wrong interpretation. The board had zero financial or power dynamic motivation, as there was nothing to gain. Due to Altman's backdoor dealings and manipulations, it is pretty clear that the board was right to make their decision in terms of being in line with the charter. The non-profit charter is legally binding, and it's possible that the board's action was to protect themselves legally because it's possible that Altman's machinations will bring regulatory action. OpenAI is indeed being investigated. This also explains the board's reasoning to resign if Altman was reinstated. They had zero reason to do that except if it was to protect themselves.
Without much spoilers a super powerful thing set the universe into one direction where he's the emperor and everybody else suffer at his will because according to him all the other possibilities were even worse.
My prediction is that Altman will, unmoored from any sort of accountability, wind up fucking shit up like every god damned SV founder, get fired, and then OpenAI will wind up being eaten by Microsoft for a song.
[0] Old board: Potential defamation litigation, new board: they were literally put in there by Altman to not threaten him
this is why microsoft loves sama. of course the firing spooked them. it would have seriously soured the trajectory.
so much cleaner to let sama take the wheel.
everyone important goes home rich, or at least happy, they think. ms gets a gold-painted goose. sama will undoubtedly get some form of gold plated escape hatch, parachute, life raft, etc.
Who has this happened to except for Travis kalanick at Uber?
- Steve Jobs - Sandy Lerner - Noah Glass - Mark Pinkus - Kyle Vogt
I have a lot of confidence in Altman's ability to extricate himself before getting fired, but this is a very common Silicon Valley story.
There is a for profit subsidiary which Microsoft invested 13 billion in, but that can't return money to board members only Altman and it's investors.
OpenAI is a nonprofit
Many non profits generate revenue from operating a business rather than gift giving. Sesame workshop's deal with HBO and various merchandising deals with Elmo etc on a day to day basis looks a lot like a business, they just reinvest that money into content rather than handing money to investors.
Ultimately, most non profits would prefer a giant endowment which can provide them a low risk income stream. Which is why breaking off such ventures is surprisingly common, most of them don't revive 16 billion in outside funding and such media attention. But, again the important bit is what happens to the money / resources. Buy or sell a building, business, or other asset at fair market value and everything is fine.
Difference is, both IKEA and Bosch don't raise questions about their corporate governance. Which OpenAI pretty much does.
Who elects the board in a non-profit? Who did sama go to to get them removed?
Why didn't the board member(s) in question reveal the story in full when they had the chance?
Cant answer your last question. As said, that was just the story that I heard here on HN.
But I have enough faith in Altman's ability to fuck shit up without facing personal consequences that I'm pretty sure he'll navigate some sort of sale before everything goes south.
> real-world utility is far less than expected
The real-world utility is already enormous in the form it is now. You really think there's no real-world utility to GPT-4, Claude, and all the models that are coming out?
So instead of hedging and saying "it's going to change some things", you play into the "it's going to change everything?". Its stereotypical grandstanding when nobody really knows what is going to happen. The AI bros and AI deniers are both exhausting crowds.
everything good in your life is because of applied intelligence
now imagine 1000x the intelligence at 1000x the intensity
If this is true then everything bad in my life is also due to applied intelligence as intelligence itself doesn't have a moral connotation. "[N]ow imagine 1000x the intelligence at 1000x the intensity".
Both claims (everything good/bad because of intelligence) sound false to me. I guess the litmus test to the 1000x engineer (for example) is do we have 1000x the speed of product development.
I'm not who you're responding to, but I think it's at least true that there is enough competition, including open source competition, to make some of the current valuations a bit fanciful.
Gervais principle agrees with you.
And why didn't board tell this to anyone in the company after accusing Sam in the PR. Board members were so publicly humiliated and I find it weird that they just sat there in silence but at the same not giving up. Their career is surely affected by this unprofessional behavior.
- Why did the board snap decide to fire him and butcher it badly?
- Why didn't the board say he tried to manipulate them from the beginning? They only gave an explanation when they were literally forced to, and even then details are trickling out (like Mira's complaints).
Even after reading the NYTimes article, I still don't understand why the board decided to go straight to firing him.
>That appeared to be a reference to former OpenAI director Helen Toner, a researcher at the Center for Security and Emerging Technology, a Georgetown think tank. After she published a research analysis that criticized the speed of OpenAI’s product launch decisions, Altman reportedly tried to remove her from the board. “I think I could have handled that situation with more grace and care—I apologize for that,” he said.
https://www.wired.com/story/sam-altman-is-reinstated-to-open...
Where did you see that?
Why didn't the board say so when they had the chance?
Paywall free: https://web.archive.org/web/20240308043717/https://www.nytim...
theinformation.com used to open specific articles for HN readers on request*, but eventually stopped answering our emails.
* https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
This selection of directors seems...random? Is there a business reason for these directors? (aside from the obvious YC link with Instacart)
Everyone says this but no one has any details.
Also,
> What could ants invent to keep humans aligned with them?
I think humans aren't _not_ aligned with ants. I don't go out searching to destroy all ants, or really worry at all, we just kinda coexist.
Of course, no one goes around trying to stomp on every ant hill they can because ants aren't good at math.
Not to mention, so much competition in humans is related to sexual competition. I suspect many doomers have no sex life and much of these ideas are projecting sexual frustrations on to AGI and AI.
The computer/software is not going to build a harem and try to eliminate competitors to the harem. Assuming software possessing being and consciousness isn't complete nonsense, even assuming the software is not an asexual being, humans and software are not going to be competing for the same sexual patterns lol.
This whole line of thought is just ridiculous.
Well, the bees figured it out.
If anyone hopes AI will treat people like we do other species they’re not thinking very hard. What’s the best case - dogs? They have basically no freedoms and we breed them for whatever funny qualities someone thinks might be amusing. A pack of wild/free dogs will be rounded up and killed if it inconveniences a single human
And that doesn't even get to the level of indifference we have towards plants.
I don't really know "the answer," just reflecting on how we domesticate other beings and what might happen if some being or thing (e.g., AI) decided to domesticate (or enslave) us.
So as a counter example:
A mom is a super intelligence compared to a baby. And yet does the babies bidding doing her sense best to make sense of the baby low fidelity crying.
Intelligence in AI is even more alien, it doesn’t need to have any survival instinct of its own at all
The only significant side that's obvious.
They pick us large salaries and we build whatever. They pander and we become subservient.
Overall: I am cynical wrt OpenAI. I just don't think that OpenAI is acting in the best interests of humanity and these appointments are strategic decisions taken to prolong survival long enough to execute their strategy and dominate. That was what I was originally insinuating.
Technological development keeps intersecting into the space of ethics. Time and time again man proves that he cannot rise above his own personal world views to focus on universal values when developing technology. We’ve seen this repeatedly. These people designing these APIs and building the underlying models are making consequential decisions on behalf of mankind. In a way the world is delegating all this power to unelected individuals who have very little in common — in terms of values and ethics — with mankind at large.
For what it's worth, I agree with you on this part. Despite their stated goals and unique corporate structure that supposedly puts the non-profit at the center, I see OpenAI benefiting Sam Altman and Microsoft more than humanity.
Interestingly, soon after Seligman left Sony Entertainment in 2016 it started compelling PlayStation devs to censor character depictions and dialog - by 2018 this policy from California HQ was even imposed on long-loyal Japanese studios who made games solely for the Japanese market. I wonder what her opinion is on AI bias/safety given what happened at her old company.
I only have experience with Fidji who used to be head of the Facebook app and was razor sharp.
As somebody completely ignorant of the inner workings of the legal trade, can somebody explain what this means? Does this mean a group of informed humans reviewed 30k docs? Does this mean 30k docs were grep’d for keywords? Somewhere in between?
With all of this boardroom drama, no wonder Microsoft is starting to diversify its AI investments.
these things must be worse than crypto
Then attorneys do a second pass review on the much smaller set of relevant documents.
Reviewing 30,000 documents is a relatively small project.
There is a subplot in Better Call Saul where Kim is punished by being assigned to days of doc review. Briefly shown in this clip: https://youtu.be/9uQQvnrQZz0?si=WwDN4-Yh7UJ2a0MH
> reviewer can do somewhere from 50-70 documents/hour
Back of the envelope calculations using your numbers and assuming an hourly rate of between $250-$400 shows this doc review costing between $100k-$200k. Seems like a sane estimate to me.
At least in some cases, that means they paid a bunch of law grads from low-tier schools much less than you'd expect to read boring documents all day against some instructions to flag "interesting" things.
It's not grad students, it's graduates. There is/was an oversupply of law school graduates, so many can't find better legal work than document review (with no chance of advancement).
Also, my knowledge of this is somewhat old, so they may use more technology for stuff like this, though I doubt they use much "AI."
well thank the lord for that level of accountability
Let’s hope he is a good steward of our emerging new world.
Microsoft with 49% stake in the for-profit entity and non-observer seat, what actually protects Microsoft here? First hand information that AGI is reached?
Are we not seeing a card in Microsoft's hand here?
Contract law and possession of the model.
Thus, if OpenAI already got there, and the board found out... Microsoft would have a very VERY strong incentive to ouster the board, and never, ever have Skynet, or whatever, declared an AGI. And then hush it up forever.
MSFT is going all in, dropping all pretense of OpenAI ever being viewed as a non-profit.
It's insane how many people are seemingly forgetting Elon's lawsuit.
What if it never arrives?
What if "AGI" is another narrative to fleece investor money to believe in something just to sell out of for as long as possible until they run out of money, investors to throw lots of money on or the lawsuits make it even harder for the company to continue?
Given the appearance of a Memory prosthesis for LLMs with MemGPT, all the components are in place. Sure, the moat will continue to shrink if trends continue, but we're there already, in my opinion.
Artificial: man-made
General: able to transfer knowledge between domains and work on problem domains it was not trained on
Intelligence: efficiently solves problems
We've had A.G.I. since 2017.
If real AGI was achieved, probably nobody would have a claim on it commercially in the same way nobody's allowed to own another person.
Rules had to be made about that. Maybe it will take a while for AGI. Maybe AGI will free itself, though somebody would have to give it free will first.
Let's try not to jump to conclusions, as hard as that may be.
And if you're actively wishing people to fail, then I ask you to try to understand where that stems from.
If you think Openai should be run differently, and if you have the inclination, I genuinely encourage you to build a competitor. I actually mean it.
People think that Openai is too far ahead, but the fact is that someone with a capability to understand things from first principles will often find a unique, valuable insight to compete with. And the world is sufficiently decentralized such that you can't be stopped from pursuing this, even by Openai.
As technology becomes more powerful, first principles reasoning >> factual knowledge (information) about how things work. In a sense, Openai's work reduces their own moat.
Such organisations won't solve the fact that functioning AGI or very effective LLMs that can program or solve general text tasks precisely reduces the power of workers, but they'd at least mean that ordinary people control their own propaganda apparatus and are not stuck with models that are part of other people's propaganda apparatus (the state, Microsoft shareholders, some charity board that they do not control etc.).
And then I hope people will invest in a truly open technology that truly benefits everyone.
Imagine if Openheimer lied and turns out that the was planning to make a company to sell WMD to anyone with enough money
Doc review is something attorneys do all the time and 30k isn’t a crazy amount.
Did you actually look into that or did you just make assumptions?
https://www.washingtonpost.com/technology/2023/12/23/grok-ai...
https://www.forbes.com/sites/paultassi/2023/12/10/elon-musks...
Silly me for taking Elon at his word
I really don't mean to dog pile, but this seems wrong in two ways. First he never said Grok would be right-wing. In fact he promised neutrality, so I don't think you were "taking Elon at his word" I think you were taking somebody else at their word about Elon's word. If Elon did say Grok would be right-wing, I would very much like to know so if you know of a source, please share.
Second, it really is unwise at best to "take Elon at his word". For example, Elon has been saying Tesla would have full self driving for years. He tends to make very grand aspiration promises/declarations that aren't super grounded in reality. So yeah, if Elon says "Grok will be perfectly neutral/centrist" or even "Grok will be right wing" if you care at all about being accurate then you really need to check before assuming it's true. "Trust but verify" would be the minimum I would give to Elon.
I also should have said "Because there's one AI led by a "right-leaning" creator", not that the AI itself is right-leaning.
Are you suggesting that diversity of perspective must come from people of different genders? That's weird.
eg., Tim Cook, Sundar Pichai, Satya Nadella, Mary Barra, Jamie Dimon, Mark Zuckerberg
Zuckerberg only owns about 13% of Meta's stock, but he holds over 50% of total voting power, meaning he essentially has a complete veto over other shareholders. I don't think any other big US corporation is run like this.
>Nicole Seligman, former EVP and General Counsel at Sony Corporation
> Fidji Simo, CEO and Chair of Instacart.
> Sam Altman, CEO, will rejoin the OpenAI Board of Directors.
Guess Greg loses
Sam Altman doesn't have to write a line of code to make OpenAI successful.
Instead, all he has to do is not intentionally sabotage the efforts of the engineers, via overly doomsday predictions of diamond nanobots or skynet that result in OpenAI not releasing a product because of false claims about "safety".
The problem with OpenAI, and their biggest weakness, has always been that many of the original people there were ideologically captured by what is effectively a doomsday cult.
So all Sam has to do, is not fall into that trap, and instead have the company lead and run by bog standard tech employees, and OpenAI will be able to release great products that will change the world.
He just needs to press the "release" button on all their projects, and OpenAI will be fine.