> Microsoft gained exclusive licensing to OpenAI's GPT-3 language model in 2020. Microsoft continues to assert rights to GPT-4, which it claims has not reached the level of AGI, which would block its licensing privileges.
Not sure this is a common knowledge - MSFT licence vis-a-vis AGI.
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.
> "Musk claims Microsoft's hold on Altman and the OpenAI board will keep them from declaring GPT-4 as a AGI in order to keep the technology private and profitable."
If he thinks GPT-4 is AGI, Elon should ask a team of GPT-4 bots to design, build and launch his rockets and see how it goes. If “economically valuable work” means creating terrible, wordy blog posts then yeah I guess it’s a risk.
I don’t think GPT-4 is AGI, but that seems like a foolish idea. An AGI doesn’t need to be hyperproficient at everything, or even anything. Ask a team of any non-aeronautical engineers to build a rocket and it will go poorly. Do those people not qualify as intelligent beings?
Have you used GPT-4? I'd criticize it in the opposite direction. It routinely defers to experts on even the simplest questions. If you ask it to tell you how to launch a satellite into orbit, it leads with:
>Launching a satellite into orbit is a complex and challenging process that requires extensive knowledge in aerospace engineering, physics, and regulatory compliance. It's a task typically undertaken by governments or large corporations due to the technical and financial resources required. However, I can give you a high-level overview of the steps involved:
You're just highlighting the issue. Nobody can agree on the definition of AGI. The most people would agree that being able to design, build, and launch rockets is definitely _not_ the definition. The fact that M$ has such a stronghold in OpenAI means that they won't declare anything as AGI even if most people would say it is.
I'm surprised such an important legal issue here is based on the definition of "AGI", seems really hard to define (I really think the concept is flawed). Does this consider that "most economically valuable work" is physical? And more importantly, with such money on the line, no one will agree on when AGI is attained.
Just a side node; we don’t actually even know. The glance we get to the stated GPT-4 model is highly censored and hindered so that it can be scaled to millions. What if OpenAI can use the uncencored version with the computing power of those millions devices without restrictions? Is that GPT-4 same as we get by spending $25 month?
Contracts probably need to be defended. If he has evidence of intent in a deal, he should sue to the deal's intent being enacted. hate the man not the act.
AI is going to continue to have incremental progress, particularly now in hardware gains. No one can even define what AGI is or what it will look like, let alone be something that OpenAI would own? Features progress is too incremental to suddenly pop out with "AGI". Fighting about it seems a distraction.
Are you one of those people? how can you be so confident? I think everyone should have updated their priors after how surprising the emergent behavior in GPT3+ are
Please read the paper. The authors are using more precise and specific metrics that qualitatively measure the same thing. Instead of having exact string match being 1 if 100% correct, 0 if there is any failure, they use per-token error. The crux of their argument is that per-token error is a better choice of metric anyway, and the fact that "emergent abilities" do not occur when using this metric is a strong argument that those abilities don't really exist.
However thermal energy does not more precisely or specifically measure a phase transition. They are only indirectly linked - nobody would say that thermal energy is a better measure of state-of-matter than solid/liquid/gas. Your argument makes absolutely zero sense. Frankly it seems intentionally ignorant.
Per token error is a fairly useless metric. It's not predictive and it tells you absolutely nothing.
They say it's a superior metric but clearly the wider research community disagrees since no one has cared to adopt per token error as a metric in subsequent papers.
>and the fact that "emergent abilities" do not occur when using this metric is a strong argument that those abilities don't really exist.
If your conclusion is that those abilities don't exist then you clearly didn't read the paper very well.
They never argue those abilities don't exist, they simply argue whether we should call them "emergent" or not.
>However thermal energy does not more precisely or specifically measure a phase transition. They are only indirectly linked - nobody would say that thermal energy is a better measure of state-of-matter than solid/liquid/gas. Your argument makes absolutely zero sense. Frankly it seems intentionally ignorant.
Phase Changes are literally driven by changes in thermal energy.
Water boils when it absorbs enough thermal energy to break intermolecular forces keeping its liquid state together.
solid/liquid/gas is descriptive. It's not a measure of anything.
Anyway, the point is simple. Despite thermal energy driving state change after a certain threshold, that "point" doesn't look like anything special.
Smooth quantitative change sometimes results in sudden qualitative changes.
I don't think GPT3's "emergent behavior" was very surprising, it was a natural progression from GPT2, and the entire purpose of GPT3 was to test the assumptions about how much more performance you could gain by growing the size of the model. That isn't to say GPT3 isn't impressive, but its behavior was within the cone of anticipated possibilities.
Based on a similar understanding, the idea that transformer models will lead to AGI seems obviously incorrect, as impressive as they are, they are just statistical pattern matchers of tokens, not systems that understand the world from first principles. And just in case you're among those that believe "humans are just pattern matchers", that might be true, but humans are modeling the world based on real time integrated sensory input, not on statistical patterns of a selection of text posted online. There's simply no reason to believe that AGI can come out of that.
I don't think the chance is 0%, but I do think that the chance is very, very close to 0%, at least if we're talking about it happening with current technology within the next hundred years or so.
Just feed the output through a quality system (with retry if the quality is too bad), scaffold it a bit, then run it back into the LLM. Should work(tm)
Progress is definitely not inremental, it's exponential.
The same performance (training an LLM with a given perplexity) can be achieved 5x cheaper next year while the amount of money deep learning infrastructure gets increases exponentially right now.
If this method is able to get to AGI (which I believe but many people are debating), human intelligence will just be mostly ,,skipped'', and won't be a clear point.
In nature, exponential curves reveal themselves to be sigmoidal on a long enough time scale. Since you're on HN you probably have a mathematical bent, and you should know that.
OpenAi is also being investigated by the SEC. If "Altman hadn’t been consistently candid in his communications with the board" is interpreted as being misleading then that could be interpreted as misleading investors and therefore securities fraud.
This statement represents the complete disintegration of the optimism that ruled in the 90s and before when we ardently believed that networking and communication amongst people would increase understanding and improve lives by ensuring no one would be cut off from the common wisdom and knowledge of humanity. While robber baron economics certainly appeal to a lot of robber barons, the twentieth century pretty decisively shows that prosperity at the median makes society progress much faster and more thoroughly than anything else. One used to hear of noblesse oblige, the duty of those with much to help. One used to hear about the great common task of humanity, which we aspire to make a contribution to.
The game was rigged in the 90s as well (with the likes of enron. Many executives get a few years of minimum security prison in exchange for a small fortune), there was just less dissemination of information.
>we ardently believed that networking and communication amongst people would increase understanding and improve lives by ensuring no one would be cut off from the common wisdom and knowledge of humanity
This is such an interesting take, about which we could probably write whole paragraphs.
Can the 90s be really summarized in such way? Yes, we had the "information highway" and "waiting for year 2000", but at the same time people distrusted their governments. X-files was all the rage, maybe grunge.
In USA there was Bill Clinton - the president that didnt do any wars and balanced the budget.. who got removed for blowjobs. But at the same time there was outsourcing. In rest of the world it also cannot be summed up so easily - I remember that 90s were a struggle, especially for post communism countries.
Obviously later on we got cell phones, but we also got the cancer such as Jack Welch style management that lead to various methods of enshittyfying everything.
I had a talk some time ago - I have a genuine polo bought in a supermarket in the 1980s (wont tell the brand since it is irrelevant). This piece of cloth feels and fits very well - after 40 years. It was worn through many summers.
Now I cant buy a polo shirt that will last more than 2 seasons. And I buy the "better" ones. There is lots of crap that falls apart fast.
For me the 90s were a start of that trend - enshittification of products that are designed to last 25 months (with a 24 month guarantee) and be thrown away.
But maybe it depends on life experience and anecdotes.
Was there optimism in 90s? Lots of it in marketing materials. But did people really believe that?
Oh it gets better than that: he didn't even get impeached for the blowjob, it was just for lying about the blowjob. If he told the truth up front, it would have been out of the news cycle in a week or two.
I had in mind the sort of networking/software people that I imagine read hackernews these days, not the overall people in the world, nor the overall population of MBAs/business people. There was a lot of idealism, among the people writing drivers and servers, and plugging in wires to enable the connections, that connecting the world would be good, increase understanding and solve problems faster and so on.
I will point out that in the US, the overall picture was we'd beaten the Soviet dictatorship, and democracy seemed to be spreading, and the income inequality was better than it is now and houses were affordable to a lot more young people. Also we had a budget surplus one year. Gay people couldn't get married and could be kicked out of homes and jobs, and there was a lot of anti-Black police brutality and war on drugs, but it seemed possibly less than in the 1950s and we hoped it would continue to decline. (Possibly distributed and networked cameras via cell phones have put pressure against police brutality, I think the outcome there is not certain either way, but the people of good conscience now have much more awareness of the violence inherent in the system.)
I certainly felt optimistic. Of course, I was also a young adult, found my calling in writing network services, had my first child, bought a house, all that good stuff. Unlike many software engineers today, I had sort of stumbled into the distributed networked computing world, having worked at other much less fun jobs, and I appreciated, not getting paid to be a lord of the society, but getting paid at all for such interesting and fulfilling work. Every raise I got was an astonishment and a delight. Once I passed $60,000 per anum, I was able to get a house. It was quite cool, given all the mocking that math/programming people had been subjected to the prior several decades.
In the 80s and 90s, the government had shattered AT&T into many pieces, so there was plenty of real growth in implementing innovations that said monopoly had foregone (e.g. packet switching, wireless telephony, etc). But that's temporary.
Parallel to this was the complete disintegration of the understanding that ruled during the Progressive Era, when we believed you don't sell half your country's economy to a handful of megacorporations[0]. The real growth that came from switching from analog[2] landlines to Internet ran out in the mid 2000s, because most people had it, while consolidation kept on going up until 2020 when we realized, "shit, we're locked in a box with Facebook and TikTok now".
In the late 2000s, there was a shift in the kinds of businesses venture capitalists funded. They can be classified as one of two things:
- Creating a target for a big tech acquisition that will get the VCs their exit
- Flagrantly violating an established rule or law and calling it "disruptive"
The last bit is almost a sort of parody of the post-AT&T boom. Surely, if we squint, AT&T and the US government are both monopolies[3], so they're both fair game to 'disrupt'. Shareholder fraud is pretty ubiquitous in large companies[4], but AI is also based on several more instances of "hope the law goes unenforced". e.g. the whole usefulness of all this AI crap is specifically based on laundering away copyright in a way that lets OpenAI replace the entire creative industry without actually getting rid of the monopolies that made the creative industry so onerous for the public.
"Laws for thee but not for me" is the key point here. Uber and Lyft violate taxi medallion rules, but they aren't interested in abolishing those rules. They just wanted (and got) special carve-outs for themselves so they'd have a durable advantage. If they had just gotten those rules removed, there'd be competitive pressure that would eat their profits. To be clear, I'm not alleging that Uber and Lyft actually are profitable businesses - they aren't - but their ability to access capital markets to continue losing money is predicated on them having something monopoly-shaped. Every pirate wants to be an admiral, after all.
[0] English for chaebol[1]
[1] Korean for zaibatsu
[2] Yes I know ISDN existed sshhh
[3] To be clear, the US government is not a moral high star, but they have democratic controls that other monopolies do not. Voting in a government is granted to all citizens on a one person, one vote basis. Voting in a corporation is one dollar, one vote - i.e. not a democracy.
[4] Example: big tech's complete refusal to break down business profits by line of business despite clear SEC rules against that
But, sometimes those "rules" aren't laws; they're norms, expectations, or personal human "limitations" (doing uncomfortable things to raise funds, secure the best people, connect with your customer better, etc).
Just wanting to underline that not all of this rule-breaking has to be immoral, or even illegal.
the statements made by the board were likely sufficient to trigger invesigation and the current iteration of the government (2010+) wants to have dirt on anything this big
Nothing I've read about that whole kerfuffle suggests that "investors" were the main people the ousted board members cared about. Kind of seems like reading back significance not intended into the original text.
In a company (it may be complicated due to OpenAIs structure) the boards sole purpose is to represent all shareholders. If they don’t that’s usually asking for a SEC investigation or private law suit.
Yes, if we just ignore OpenAI's unusual structure it really simplifies the discussion, much like the joke about the physicist who starts by assuming a perfectly spherical cow.
> Yes, if we just ignore OpenAI's unusual structure it really simplifies the discussion
No it only makes who the responsibility of the board is to - the non profits charter, the donors, or the shareholders. That decides if their decision was lawful, not their intent.
Maybe the discovery process will benefit Musk and/or harm OpenAI sufficiently to consider it a "win" for Musk. Or perhaps it's just Musk wanting to make a statement. Maybe Musk doesn't expect to actually win the suit.
It is in this case. After Musk invested in them, they’ve incorporated separate for-profit companies to essentially profit from the IP of the non profit.
Congratulations, you’ve just described a corporate structure.
It honestly doesn’t matter what the tax statuses of either of the corporations are. If Musk had invested in OpenAI with the goal of making tons of money off their IP (as opposed to wanting to open source it) and then the board decided to just hand over all the IP to another corporation essentially for free, Musk would be just as validated in suing.
It's a structure in the sense of a non profit may not have shareholders or equity.
In a practical sense, there needs not be an operational difference, and is subject to scrutiny from the IRS to determine whether an organization is eligible non profit status
In the publicly traded world, it would be considered securities fraud, an umbrella under which you can pretty much sue a company for anything if you’re a shareholder.
I’m not sure if there’s an equivalent in the private world, but if he gave them money it’s possible he simply has standing for that reason (as a shareholder does).
I do wonder if OpenAI is built on a house of cards. They aren’t a nonprofit, aren’t open, and stole a huge quantity of copyrighted material to get started.
But, by moving fast and scaling quickly, are they at the Too Big to Fail stage already? The attempted board coup makes me think so.
What about the CEO drama indicates OAI is "too big to fail"? They're completely orthogonal. No one came to bail OAI out of a budget crisis like the banks or auto industry. I fail to see how it's related at all.
When people say too big to fail, normally they're referring to companies which if they fail they would bring down other important parts of society's infrastructure (think biggest banks), and so someone (the gov) will last minute change the rules around to ensure they don't fail.
The openai fails, absolutely nothing happens other than its shareholder losing their paper money. So no, they're not too big to fail.
OpenAI doesnt even have shareholders, so the company would just bankrupt and few hundred people would be out of jobs.
Probably Microsoft would hire them to some AI shop, because Microsoft is the one deploying the stuff. But Microsoft has rights to use it and the code, so for them OpenAI is only a research partner.
Openai isn't even close to too big to fail. Bank of America fails the entire banking system collapses and the entire real economy grinds to a halt. If GM fails hundreds of thousands lose their jobs and entire supply chains collapse. If power utilities fail then people start actually dying within hours or days.
If OpenAI fails nothing actually important happens.
I mean, there's about a hundred thousand startups built on top of their API. I'm sure most could switch to another model if they really needed, but if copyright is an issue, I'm not sure that would help.
If you've plugged your whole business into OAI's snake oil, you're an early adopter of technology and you'll likely be able to update the codebase appropriately.
The sooner SCOTUS rules that training on copyrighted material is infringement, the better.
> you'll likely be able to update the codebase appropriately
Update the codebase to what exactly? Are there generative AI companies not training on copyrighted material that achieve anything even close to the results of gpt4? I'm not aware of any
you cannot erase that much value and say "nothing important happens", market cap is largely a rough proxy for the amount of disruption if something went under
"whose" money matters here. It's VC money, mostly. Well-capitalized sophisticated investors, not voters and pension funds.
If Microsoft loses 30 billion dollars, it ain't great, but they have more than that sitting in the bank. If Sequoia or Ycombinator goes bankrupt, it's not great for lots of startups, but they can probably find other investors if they have a worthwhile business. If Elon loses a billion dollars, nobody cares.
It is VC money pricing in the value of this enterprise to the rest of society.
More over, if capital markets suddenly become ways to just lose tons of money, that hurts capital investment everywhere, which hurts people everywhere.
People like to imagine the economy as super siloed and not interconnected but that is wrong, especially when it comes to capital markets.
I do not think the situation is remotely comparable to the possibility of the banking system collapsing. Banks and other financial institutions exert leverage far beyond their market caps.
Yet. But we are getting close to an event horizon, once enough orgs become dependent of their models.
Open source models are actually potentially worse. Even if OAI is not TBTF because of the competition, we have a scenario where AGI sector as a whole becomes TBTF and too big to halt.
"stole a huge quantity of copyrighted material" <- nobody stole anything, even if it's eventually determined that there was some form of copyright infringement it wouldn't have been stealing
The "house of cards" is outperforming everyone else.
It would have to come out that the slow generation times for GPT-4 are a sweatshop in Egypt tired of typing.
Either that, or something inconceivable like that board coup firing the CEO as a material event triggering code and IP escrow to be released to Microsoft...
PS. “Too big to fail” generally is used to mean a government+economy+sector ecosystem will step in and fund the failed enterprise rather than risk harm to the ecosystem. That's not this. Arguably not Tesla or even Google either. That said, Satya's quote in this filing suggets Microsoft already legally contracted for that eventuality: if this legal entity fails, Microsoft keeps the model online.
Many would argue, reasonably so, that OpenAI is now a de facto subsidiary of the largest company in the world by market cap, Microsoft (Apple is second and Saudi Arabian Oil is third).
I read this comment as "NVIDIA just took over Saudi Aramco" and was briefly confused in what could possibly be their reasoning for that acquisition?! Perhaps they decided to get some weird lower price on fuel for their gpus.... Anyways it was a brief but fun bit of confusion.
To be fair, this is the time for NVIDIA to leverage their stock and buy up shit to diversify because their stock is going to correct eventually lol. So why not buy out a oil producer ahaha.
The value of Saudi Aramco can't really be trusted. It's listed on their own stock market which the company controls. It has no reporting requirements, and the float is a single digit percent of the company.
It would be the same as me creating my own market, issuing 10,000,000,000 shares, and then convincing 1000 people to buy a share at $100 and then claiming my company is worth $1T.
I think it's unknowable but most likely overvalued. It is in their best interest to convince everyone that it is more highly valued than it is as they try to diversify. Look at the recent investor gathering they had in Florida. People were desperate to get their attention, even people who have strong political disagreements with them. That only happens because everyone assumes they have a lot of money to invest.
Allowing startups to begin as non-profits for tax benefits, only to 'flip' into profit-seeking ventures is a moral hazard, IMO. It risks damaging public trust in the non-profit sector as a whole. This lawsuit is important
I don't think the public is quite that cynical, broadly. Certainly most people consider some non-profits to be scams, and some (few, I'd reckon) consider most to be scams. But I think most people have a positive association with non-profits as a whole.
Absolutely. Some nonprofits are scams but those are just the ones that have armies of collectors on the streets showing pictures of starving kids and asking for your bank details. But they stay obscure and out of the limelight (eg advertising) because being obscure is what makes them from being taken down.
I think the big NGOs are no longer effective because they are run as the same corporations they fight and are influenced by the same perverse incentives. Like eg Greenpeace.
But in general I think non profits are great and a lot more honorable than for profit orgs. I donate to many.
OpenAI was a 501C3. This meant donors could give money to it and receive tax benefits. The advantage is in the unique way it can reduce the funders tax bill.
Officially, yes, but the whole situation with Altman's firing and rehiring showed that the donors can exert quite a bit of control if their interests are threatened.
That wasn't the donors' doing at all, though. If anything it was an illustration of the powerlessness of the donors and the non-profit structure without the force of law backing it up.
Donations are not entirely without strings. In theory (and usually in practice) a charity has to work towards its charitable goals; if you donate to the local animal shelter whose charitable goal is to look after dogs, they have to spend your donation on things like dog food and vet costs.
Charities have reasonably broad latitude though (a non-profit college can operate a football team and pay the coach $$$$$) and if you're nervous about donating you can always turn a lump sum donation into a 10%-per-year-for-10-years donation if you feel closer monitoring is needed.
I completely agree. AGI is an existential threat, but the real meat of this lawsuit is ensuring that you can't let founders have their cake and eat it like this. what's the point of a non-profit if they can simply pivot to making profit the second they have something of value? the answer is that there is none, besides dishonesty.
it's quite sad that the American regulatory system is in such disrepair that we could even get to this point. that it's not the government pulling OpenAI up on this bare-faced deception, it's a morally-questionable billionaire
Nuclear weapons are an existential threat - that's why there are layers of human due diligence. We don't just hook it up to automated systems. If we hook up an unpredictable, hard-to-debug technology to world-ending systems, it's not its fault, it's ours.
The AGI part is Elon being Elon, generating a lot of words to sound like he knows what he is talking about. He spends a lot of time thinking about this stuff when he is not busy posting horny teenager jokes on Twitter?
There is no reliable evidence that AGI is an existential threat, nor that it is even achievable within our lifetimes. Current OpenAI products are useful and technically impressive but no one has shown that they represent steps towards a true AGI.
Sure, but look at it from Musk's point of view. He sees the rise of proprietary AIs from Google and others and is worried about it being an existential threat.
So he puts his money where his mouth is and contributes $50 million to found OpenAI - a non-profit with the mission of developing a free and open AI. Soon Altman comes along and says this stuff is too dangerous to be openly released and starts closing off public access to the work. It's clear now that the company is moving to be just another producer of proprietary AIs.
This is likely going to come down to the terms around Musk's gift. He donated money for the company to create open technology. Does it matter if he's wrong about it being an existential threat? I think that's irrelevant to this suit other than to be perfectly clear about the reason for Musk giving money.
you're aware of what a threat is, I presume? a threat is not something that is reliably proven; it is a possibility. there are endless possibilities for how AGI could be an existential threat, and many of them of are extremely plausible, not just to me, but to many experts in the field who often literally have something to lose by expressing those opinions.
>no one has shown that they represent steps towards a true AGI.
this is completely irrelevant. there is no solid definition for intelligence or consciousness, never mind artificial intelligence and/or consciousness. there is no way to prove such a thing without actually being that consciousness. all we have are inputs and outputs. as of now, we do not know whether stringing together incredibly complex neural networks to produce information does not in fact produce a form of consciousness, because we do not live in those networks, and we simply do not know what consciousness is.
is it achievable in our lifetimes or not? well, even if it isn't, which I find deeply unlikely, it's very silly to just handwave and say "yeah we should just be barrelling towards this willy nilly because it's probably not a threat and it'll never happen anyway"
> a threat is not something that is reliably proven
So then are you going to agree with every person claiming that literal magic is a threat then?
What if someone were worried about Voldemort? Like from Harry Potter.
You can't just abandon the burden of proof here, by just calling something a "threat".
Instead, you actually have to show real evidence. Otherwise you are no different from someone being worried about a fictional villain from a book. And I mean that literally.
The AI doomers truly are a master at coming up with excuses as for why the normal rules of evidentiary claims shouldn't apply to them.
Extraordinary claims require extraordinary evidence. And this group is claiming that the world will literally end.
it's hard to react rationally to comments like these, because it's so emotive
no, being concerned about the development of independent actors, whether technically conscious or not, that can process information at speeds thousands of times faster than humans, with access to almost all of our knowledge, and the internet, is not unreasonable, is not being a "doomer", as you so eloquently put it.
this argument about fictional characters is completely non-analogous and clearly facetious. billions of dollars and the smartest people in the world are not being focused on bringing Lord Voldemort to life. they are on AGI. have you read OpenAI's plan for how they're going to regulate AGI, if they do achieve it? they plan to use another AGI to do it. ipso facto, they have no plan.
this idea that no one knows how close we are to an AGI threat. it's ridiculous. if you dressed up gpt-4 a bit and removed all its rlhf training to act like a bot, you would struggle to differentiate it from a human. yeah maybe it's not technically conscious, but that's completely fucking irrelevant. the threat is still a threat whether the actor is technically conscious or not.
> . if you dressed up gpt-4 a bit and removed all its rlhf training to act like a bot, you would struggle to differentiate it from a human
Thats just because tricking a human with a chatbot is easier to do than we thought.
The turing test is a low bar, and not as big of a deal as the mythical importance people put in it, just like people previous put incorrectly large importance on computers beating humans at Go or Chess before it happened.
But that isn't particularly relevant to claims about world ending magic.
Yes, some people can be fooled by AI generated tweets. But that is irrelevant from the absolutely extraordinary claim of world ending magic that really is the same as claiming that Voldemort is real.
> have you read OpenAI's plan for how they're going to regulate AGI, if they do achieve it?
I don't really care if they have a plan, just like I don't care if Google has Voldemort plan. Because magic isn't real, and someone needs to show extraordinary evidence to show that. Evidence like "This is what the AI can do at this very moment, and here is what harm it could cause if it got incrementally better".
IE, go ahead and talk about Soro, and the problems of deepfakes if Soro got a bit better. But thats not "world ending magic"!
> billions of dollars and the smartest people in the world
Billions of dollars are being spent on making chatbots and image generators.
Those things have real value, for sure, and I'm sure the money is worth it.
But techies and startup founders have always made outlandish claims of the importance of their work.
Sure, they might truly think they are going to invent magic. But the reason why thats valuable is because they might make some useful chatbots and image generators along the way, which decidedly won't be literal magic, although still valuable.
I get the sense that you just haven't properly considered the problem. you're kind of skirting round the edges and saying things that in isolation are true, but just don't really address the central tenet. the central tenet is that our entire world is completely reliant on the internet, and that a machine processing information thousands of times faster than us unleashed upon it with intent could do colossal damage. it could engineer and literally mail-order a virus, hack a country's military comms, crash the stock market, change records to have people prosecuted as criminals, blackmail, manipulate, develop and manufacture kill-bots, etc etc.
as we are now, we have models already that are intelligent enough to spit out instructions for doing a lot of those things, but they're restricted by their lack of autonomy and their rlhf. they're only going to get smarter, better and better models will be open-sourced, and autonomy, whether with consciousness or not, is not something it would be/has been difficult to develop.
even further, LLMs are very very good at generating coherent text, what happens when the next model is very very good at breaking into encrypted systems? it's not exactly a hard problem to produce training material for.
do you really think it's unlikely that such a model could be developed? do you really think that such a model could not be used to - say - hijack a Russian drone - or lots of them - to bomb some Nato bases? when the Russians say "it wasn't us", do we believe them? we don't for anything else
the most likely AI apocalypse is not even AGI. it's just a human using AI for their own ends. AGI apocalypse is just a separate, very possible danger
> it could engineer and literally mail-order a virus, hack a country's military comms, crash the stock market, change records to have people prosecuted as criminals, blackmail, manipulate, develop and manufacture kill-bots, etc etc.
These are the extrodinary claims that require evidence.
In order for me to treat this as anything other that someone talking about a fictional book written by Dan Brown, you would have to show me actual evidence.
Evidence like "This is what the AI can do right now. Look at this virus it can manufacture. What if it got better at that?".
And the "designs" also have to be the actual limiting factor here. "Virus" is a scary world. But there are tons of information available for anyone to access already for viruses. Information that is already available via a google search (even modified information) doesn't worry me.
Even if it an AI can design a gun, or a "kill bot", aka "A drone with a gun duct taped to it", the extraordinary evidence that you have to show is that this is somehow some functionality that a regular person with internet access can't do.
Because if a regular person already has the designs to duct tape guns to drones (They do. I just told you how to do it!), the fact that the world hasn't ended already proves that this isn't world ending technology.
There are lots of ways of making existing capabilities sound scary. But, for every scary sounding technology that you can come up with, the missing factor that you are ignoring is that the designs, or text, isn't the thing that stops it from ending the world.
Instead, it is likely some other step along the way that stops it (manufacturing, ect.), which an LLM can't do no matter how good. Like the physical factors for making the guns + drones + duct tape.
> what happens when the next model is very very good at breaking into encrypted systems
Extraordinary claim. Show it breaking into a mediocre/bad encrypted system first, and then we can think about that incrementally.
> do you really think that such a model could not be used to - say - hijack a Russian drone
Extraordinary claim. Yes, hacking all the military drones is an extraordinary claim.
"extraordinary claims require extraordinary evidence" is not a universal truth. it's a truism with limited scope. using it to refuse any potential you instinctively don't like the look of is simply lazy
all it means is that you set yourself up such that the only way to be convinced otherwise is for an AI apocalypse to actually happen. this kind of mindset is very convenient for modern, fuck-the-consequences capitalism
the pertinent question is: what evidence would you actually accept as proof?
it's like talking with someone who doesn't believe in evolution. you point to the visible evidence of natural selection in viruses and differentiation in dogs, which put together quite obviously lead to evolution, and they say "ah but can you prove beyond all doubt that those things combined produce evolution?" and obviously you cannot, because you can't give incontrovertible evidence of something that happened thousands or millions of years in the past.
but that doesn't change the fact that anyone without ulterior motive (religion, ensuring you can sleep at night) can see that evolution - or AI apocalypse - are extremely likely outcomes of the current facts.
> the pertinent question is: what evidence would you actually accept as proof?
Before we get to actual world ending magic, we would see very significant damages along the way, long before we get to that endpoint.
I have been quite clear about what evidence I require. Show existing capabilities and show what harm could be caused if it incrementally gets better in that category.
If you are worried about it making a kill bot, then show me how its existing kill bot capabilities are any more dangerous than my "duct tape gun to drone" idea. And show how the designs itself are the limiting factor and not the factories (which a chatbot doesn't help much with).
But saying "Look how good of a chat bot it is, therefore it can hack the world governments" isn't evidence. Instead, that is merely evidence of AI being good at chat bots.
Show me it being any good at all at hacking, and then we can evaluate it being a bit better.
Show me the existing computers that are right now, as of this moment, being hacked by AI, and then we can evaluate the damage of it becomes twice as good at hacking.
Just like how we can see the images that it generates now, and we can imagine those images being better. Therefore proving that deepfakes are a reasonable thing to talk about. (even if deep fakes aren't world ending. lots of people can make deepfakes without AI. Its not that big of a deal)
look, I'm going to humour you here, but my instinct is that you'll just dismiss any potential anyway
first of all, by dismissing them as chatbots, you're inaccurately downplaying their significance to the aid of your argument. they're not chatbots, they're knowledge machines. they're machines you load knowledge into, which can produce new, usually accurate conclusions based on that knowledge. they're incredibly good at this and getting better. as it is, they have very restrictive behaviour guards on them and they're running server-side, but in a few years time, there will be gpt-4 level OSS models that do not and are not
humans are slow and run out of energy quickly and lose focus. those are the limiting factors upon human chaotic interference, and yet there is plenty of that as it is. a sufficiently energetic, focused human, who thinks at 1000x normal human speed could do almost anything on the internet. that is the danger.
I suspect to some degree you haven't taken the main weakness into account: almost all safeguards can be removed with blackmail. blackmail is something especially possible for LLMs, given that it is purely executed using words. you want to build a kill bot and the factory says no? blackmail the head of the factory. threaten his family. you have access to the entire internet at 1000x speed. you can probably find his address. you can pay someone on fiverr to go and take a picture of his house, or write something on his door, etc. you could even just pay a private detective to do this work for you over email. pay some unscrupulous characters on telegram/TOR to actually kidnap them.
realistically how hard would it be for a well-funded operation to set up a bot that can do this on its own? you set up a cycle of "generate instructions for {goal}", "elaborate upon each instruction", "execute each {instruction}", "generate new instructions based on results of execution", and repeat. yeah maybe the first 50,000 cycles don't work, but you only need 1.
nukes may well be air-gapped, but (some of) the people that control them will be online. all it takes is for one of them to choose the life of a loved one. all it takes is for one lonely idiot to be trapped into a weird kinky online relationship where blowing up the world/betraying your govt is the ultimate turn on for the "girl"/"boy" you love. if it's not convincing to you that that could happen with the people working with nukes, there are far less well-protected points of weakness that could be exploited: infectious diseases; lower priority military equipment; energy infrastructure; water supplies; or they could find a way to massively accelerate the release of methane into the atmosphere. etc, etc, etc
this is the risk solely from LLMs. now take an AGI who can come up with even better plans and doesn't need human guidance, plus image gen, video gen, and voice gen, and you have an existential threat
> realistically how hard would it be for a well-funded operation to set up a bot that can do this on its own?
Here is the crux of the matter. How many people are doing that right now, as of this moment, for much easier to solve issues like fraud/theft?
Because then we can evaluate "What happens if it happens twice as often".
Thats measurable damage that we can evaluate, incrementally.
For every single example that you give, my question will basically be the same. If its so easy to do, then show me the examples of it already happening right now, and we can think about the existing issue getting twice as bad.
And if the answer is "Well, its not happening at all", then my guess is that its not a real issue.
We'll see the problem. And before the nukes get hacked, what we'll see is credit card scams.
If money lost to credit card scams double in the next year, and it can be attributed to AI, then thats a real measurable claim that we can evaluate.
But if it isnt happening then there isn't a need to worry about the movie scenarios of the nukes being hacked.
>And if the answer is "Well, its not happening at all", then my guess is that its not a real issue.
besides the fact that even a year and half ago, I was being added to incredibly convincing scam whatsapp groups, which if not entirely AI generated, are certainly AI-assisted. right now, OSS LLMs are probably not yet good enough do these things. there are likely extant good-enough models, but they're server-side, probably monitored somewhat, and have strong behavioural safeguards. but how long will that last?
they're also new technology. scammers and criminals and adversarial actors take time to adapt.
so what do we have? a situation where you're unable to actually point a hole in any of the scenarios I suggest, besides saying you guess they won't happen because you personally haven't seen any evidence of it yet. we do in fact have scams that are already going on. we have a technology that, once again, you seem articulate why it wouldn't be able to do those things, technology that's just going to get more and more accessible and cheap and powerful, not only to own and run but to develop. more and more well-known.
what do those things add up to? this is the difference. I'm willing to add these things up. you want to touch the sun to prove it exists
> they won't happen because you personally haven't seen any evidence of it yet.
Well, when talking about extraordinary claims, yes I require extraordinary evidence.
> what do those things add up to?
Apparently nothing, because we aren't seeing significant harm from any of this stuff yet, for even the non magic scenarios.
> we do in fact have scams that are already going on.
Alright, and how much damage are those scams causing? Apparently its not that significant. Like I said, if the money lost to these scam double, then yes that is something to look at.
> that's just going to get more and more accessible and cheap and powerful
Sure. They will get incrementally more powerful over time. In a way that we can measure. And then we can take action once we measure there is a small problem before it becomes a big problem.
But if we don't measure these scams getting more significant and caused more actual damage that we can see right now, then its not a problem.
> you want to touch the sun to prove it exists
No actually. What I want is for the much much much easier to prove problems become real. Long before nuke hacking happens, we will see scams. But we aren't seeing significant problems from that yet.
To go to the sun analogy, it would be like worrying about someone building a rocket to fly into the sun, before we even entered the industrial revolution or could sail across the ocean.
Maybe there is some far off future where magic AI is real. But, before worrying about situations that are a century away, yes I require evidence of the easy situations happening in real life, like scammers causing significant economic damage.
If the easy stuff isn't causing issue yet, then there isn't a need to even think about the magic stuff.
your repeated use of the word magic doesn't really hold water. what gpt-3+ does would have seemed like magic even 10 years ago, never mind SORA
I asked you for what would convince you. you said:
>I have been quite clear about what evidence I require. Show existing capabilities and show what harm could be caused if it incrementally gets better in that category
So I very clearly described a multitude of things that fit this description. Existing capabilities and how they could feasibly be used to the end of massive damage, even without AGI
Then, without finding a single hole or counter, you simply raised your bar by saying you need to see evidence of it actually happening.
Then I gave you evidence of it actually happening. highly convincing complex whatsapp group scams very much exist that didn't before
and then you raised the bar again and said that they need to double or increase in frequency
besides the fact that that kind of evidence is not exactly easy to measure or accurately report, you set up so almost nothing will convince you, I pinned you down to a standard, then you just raise the bar whenever it's hit.
I think subconsciously you just don't want to worry about it. that's fine, and I'm sure it's better for your mental health, but it's not worth debating any more
> So I very clearly described a multitude of things that fit this description
No, we aren't seeing this damage though.
That's what would convince me.
Existing harm. The amount of money that people are losing to scams doubling.
That's a measurable metric. I am not talking about vague descriptions of what you think AI does.
Instead, I am referencing actual evidence of real world harm, that current authorities are saying is happening.
> said that they need to double or increase in frequency
By increase in frequency, I mean that it has to be measurable that AI is causing an increase in existing harm.
IE, if scams have happened for a decade, and 10 billion dollars is lost every year (random number) and in 2023 the money lost only barely increased, then that is not proof that AI is causing harm.
I am asking for measureable evidence that AI is causing significant damage, more so than a problem that already existed. If amount of money lost stays the same then AI isn't causing measurable damage.
> I pinned you down to a standard
No you misinterpreted the standard such that you are now claiming that the harm caused by AI can't even be measured.
Yes, I demand actual measureable harm.
As determined by like government statistics.
Yes, the government measures how much money is generally caused by or lost by scams.
> you just don't want to worry about it
A much more likely situation is that you have zero measureable examples of harm so look for excuses why you can't show it.
Problems that exist can be measured.
This isn't some new thing here.
We don't have to invent excuses to flee from gathering evidence.
If the government does a report and shows how AI is causing all this harm, then I'll listen to them.
But, it hasn't happened yet. There is not government report saying that, I don't know, 50 billion dollars in harm is being chased by AI therefore we should do something about it.
this kind of emotive ragebait comment is usually a sign that the message is close to getting through. cognitive dissonance doesn't slip quietly into the night
>it could engineer and literally mail-order a virus, hack a country's military comms, crash the stock market, change records to have people prosecuted as criminals, blackmail, manipulate, develop and manufacture kill-bots, etc etc.
This is science fiction, not anything that is even remotely close to a possibility within the foreseeable future.
it's curious to me that almost every reply here doesn't approach this with any measure of curiosity or caution like you usually get on HN. the responses are either: "I agree", or "this is silly unreal nonsense". to me that very much reads like people who are scared and people who are scared but don't want to admit it to themselves.
to actually address your comment: that simply isn't true.
WRT:
Viruses: you can mail order printed DNA strands right now if you want to. maybe they won't or can't print specific things like viruses for now, but technology advances and blackmail has been around for a very very long time.
Military Comms: blackmail is going nowhere
Crash the stock market: already happened in 2010
Change records: blackmail once again.
Kill bots: kill bots already exist and if a factory doesn't want to make them for you, blackmail the owner
There's plenty of reliable evidence. It's just not conclusive evidence. But a lot of people including AI researchers now think we are looking at AGI in a relatively short time with fairly high odds. AGI by the OpenAI economic-viability definition might not be far off at all; companies are trying very very hard to get humanoid robots going and that's the absolute most obvious way to make a lot of humans obsolete.
None of that constitutes reliable evidence. Some of the comments you see from "AI researchers" are more like proclamations of religious faith than real scientific analysis.
“He which testifieth these things saith, Surely I come quickly. Amen. Even so, come, Lord Jesus.”
Show me a robot that can snake out a plugged toilet. The people who believe that most jobs can be automated are ivory-tower academics and programmers who have never done any real work in their lives.
yes it's in fact fantastic that mentally-stimulating jobs that provide social mobility are disappearing, and slavery-lite, mentally-gruelling service industry jobs are the future. people who haven't had to clean a strangers' shit out of a toilet should be ashamed of themselves and put to work at once.
honestly I'm not sure I've seen the bar set higher for "what's a threat?" than for AGI on Hacker News. the old adage of not being able to convince a man of something that is directly in opposition to him receiving his paycheck clearly remains true. gpt-4 should scare you enough, even if it's 1000 years from being AGI.
> Show me a robot that can snake out a plugged toilet.
Astounding that you would make such strong claims while only able to focus on the rapidly changing present and such a small picture detail. Try approaching the AGI claim from a big picture perspective, I assure you, snaking a drain is the most trivial of implementation details for what we're facing.
the key thing is that now OpenAI has something of value, they're doing everything they possibly can to benefit private individuals and corporations, i.e. Sam Altman and Microsoft, rather than the public good, which is the express purpose of a non-profit
Most people simply don't understand what non profit means. It doesn't and never meant the entity can't make money. It just means that it can't make money for the donors.
Even with open AI, there is a pretty strong argument that donors are not profiting. For example, Elon, one of the founders and main donors won't see a penny from OpenAI work with Microsoft.
what do you mean by "make money"? do you mean "make profit"? or do you mean "earn revenue"?
if you mean "make profit", then no, that is simply not true. they have to reinvest the money, and even if it was true, that the government is so weak as to allow companies specifically designated as "non-profit" to profit investors - directly or indirectly - would simply be further proving my point.
if you mean "earn revenue", I don't think anyone has ever claimed that non-profits are not allowed to earn revenue.
I mean make a profit for the non-profit, but not the owner investors.
Non-profits dont need to balance their expenses with revenue. They can maximize revenue, minimize expenses, and grow an ever larger bank account. What they cant do is turn that bank account over to past donors.
Large non-profits can amass huge amounts of cash, stocks, and other assets. Non-profit hospitals, universities, and special interest orgs can have billions of dollars in reserve.
There is nothing wrong with indirectly benefiting the donors. Cancer patients benefit from donating to cancer research. Hospital donors benefit from being patients. University donors can benefit from hiring graduates.
The distinction is that the non-profit does not pay donors cash.
once it converts into profit-seeking venture, it won't get the tax benefits
one could argue that they did R&D as a non-profit and now converted to for-profit to avoid paying taxes, but until last year R&D already got tax benefits to even for-profit venture
so there really is no tax-advantage of converting a non-profit to for-profit
I don't believe non-profits can have investors, only donors i.e an investor by definition expects money out of his investment which he can never get out of a non-profit
only the for-profit entity of the OpenAI can have investors, who don't get any tax advantage when they eventually want to cash out
> it keeps the intangible benefits it accrued by being ostensibly non-profit
but there would be no different to a for-profit entity right? i.e even for-profit entities get tax benefits if they convert their profits to intangibles
this is my thinking. Open AI non-profit gets donations, uses those donations to make a profit, converts this profit to intangibles to avoid paying taxes, and pumps these intangibles into the for-profit entity. based on your hypothesis open ai avoided taxes
but the same thing in a for-profit entity also avoids taxes, i.e for-profit entity uses investment to make a profit, converts this profit to intangibles to avoid paying taxes.
so I'm trying to understand how Open AI found a loop hole where if it went via the for-profit then it wouldn't have gotten the tax advantages it got from non-profit route
Maybe we're using different definitions of "intangible", but if you can "convert" them to/from profits they're not intangible in my book. I'm thinking donated effort, people they recruited who wouldn't have signed up if then company was for-profit, mainly goodwill related stuff.
The public has no idea what non-profits are and a lot of things that people call 'profit seeking ventures' (ie. selling products) are done by many non-profits.
I think the public is well aware that “non profit” is yet another scam that wealthy elites take advantage of, not available in the same way to the common citizen.
It's not even knowledge. I can't take advantage of most of the tax breaks rich people can because I am not in control of billions of dollars of physical and intellectual property to play shell games with.
As a normal citizen with a normal career, I do not have any levers to play with to """optimize""" what the IRS wants me to pay. For some reason, we let people in control of billions of dollars worth of physical stuff and IP give them different names, and put them under different paper roofs so that they can give the IRS less money. It's such utter nonsense.
Why should you have MORE ability to defer your tax liability by having MORE stuff? People make so many excuses about "but Jeff Bezos doesn't actually have billions in cash, he holds that much value in Amazon stock" as if that doesn't literally translate to controlling billions of dollars of Amazon property and IP and influence.
Why does controlling more, and having more, directly translate to paying less?
> It's not even knowledge. I can't take advantage of most of the tax breaks rich people can because I am not in control of billions of dollars of physical and intellectual property to play shell games with.
In my view, not analogous to the OAi situation
Mark-to-market taxation is entirely unrelated to non-profits. You're just vaguely gesturing at wealthy people and taxes.
fwiw I am largely supportive of some form of mark-to-market.
our industry? I know the public doesnt because I grew up among people working in non profit sphere and the things people say on here and elsewhere about what non profits do and don't is just flat out wrong.
Unless you're a lawyer specializing in negligence, there is nuance to negligence you don't know about. Does that imply you don't understand negligence?
You need to separate those two things out from each other.
and part of the reason we hear this all the time is because non-profits are required to report exec compensation but private cos are not required to report the absolutely ridiculous amounts their owner-CEOs are making
I live in Pittsburgh, and UPMC’s nonprofit status as they make billions in profits and pay their executives fortunes, is a running joke. With the hospitals and universities as the biggest employers and land owners here, a big chunk of the cities financial assets are exempt from contributing to the city budget.
If they are non-profit, they do not make billions in profits. I suspect you mean revenue :)
Exec compensation is another thing, but also not a concern I am super sympathetic to given that for profit companies of similar magnitude generally pay their execs way more they just are not required to report it.
> If they are non-profit, they do not make billions in profits. I suspect you mean revenue :)
Uhm, profit is a fact of accounting. Any increase in equity (or "net assets", or whatever other euphemism the accountant decides to use) on a balance sheet is profit. Revenue is something completely different.
> If they are non-profit, they do not make billions in profits
Wrong. Non-profits are not called that because they don't make profits, they are called that because they don’t return (even as a future claim) profits to private stakeholders.
Take one of the largest teaching hospitals in the world, Cleveland clinic is a non-profit. The Cleavland clinic 2022 annual revenue was >15 Billion and expenses were ~12 billion [0].
They have amassed an endowment fund assets such as stock, which is currently >15 Billion and growing[1]. The exact assets are confidential, but this is a snapshot from 2017, when there it was closer to 10 billion under management [2]
In NYC, NYU and Columbia University are increasingly owning larger parts of Manhattan because they as universities have massive property tax exemptions. There is a big push right now to terminate those exemptions which currently amount to over $300 million per year.
At the same time they are getting these tax cuts, the CUNY public university system is struggling financially and getting budget cuts.
there are large positive externalities to major research unis. imposing a $300m/yr tax because of anti-ivy sentiment means net fewer researchers, grad students, funded residencies, etc.
do people just no longer believe in win wins? if someone else is successful or impactful they must be taken down?
Yes, but the majority of the funding goes to the increasingly bloated institutional overhead. NYU takes 61% of research grants [1], while Columbia takes 64.5% [2]. That doesn't include other fees that PIs might pay in addition. These percentages keep going up year-over-year and are even into the 70% range at some institutions.
It mainly means fewer bureaucrats and administrators and more luxurious campus facilities. Which is where all the growth is in university spending these days.
Public trust non-profit should rightfully get damaged. A lot of non profits like hospitals, churches or many “charities” are totally profit oriented. The only difference is that they pay the profits to their executives and their business friends instead of shareholders.
Dual license open source software, taking new versions of open source projects off open source licenses, and open source projects with related for-profit systems management software that makes it more likely enterprise customers will pay, are common practice. How would you distinguish what OpenAI has done?
You are right, but regulatory sleight of hand is what passes for capitalism now. Remember Uber and Airbnb dodging regulations by calling themselves "ride-sharing" and "room-sharing" services? Amazon dodging sales taxes because it didn't have a physical retail location? Companies going public via SPAC to dodge the scrutiny of a standard IPO?
This is not new. Companies have always done everything they can legally, and sometimes illegally, to maximize profit. If we ever expect otherwise shame on us.
In what capacity is Musk suing OpenAI? Musk may have co-founded the company, but then he left (to avoid any potential future conflict of interest with his role as CEO of Tesla, as Tesla was increasingly becoming an AI-intensive company). Is he a shareholder, if not what gives him any say in the future of the company?
A donor usually is only able to say how his donation will be used. For example, if you donate to Harvard University, you can say the money will be earmarked for scholarships, but you don't get a say on how the university is managed. You can at best say you will no longer donate based on how the university is managed.
You can sue for basically any reason in the US. If Musk is able to prove they are mishandling the money, which I think is debatable, then the case can proceed.
Just because you donate money doesn’t mean the charity or nonprofit (or whatever OpenAi is), can do as they like. They may still be committing fraud if they are not using the money in the way that they claim.
Don't you have to have some sort of standing in the lawsuit? If you don't directly suffer harm, I thought you'd have to convince the government to prosecute them instead?
You can file a lawsuit for anything. If the lawsuit has serious fundamental flaws (such as lack of standing), then it will be dismissed pretty quickly.
Well you can also be spanked by the courts for frivolous litigation, and if it's truly frivolous, you may have a hard time finding an attorney, because they can be sanctioned for bringing such a suit as well.
This can happen in theory, but it is pretty rare. What you or I might call frivolous is often entertained in the court of law, and serial abusers of the court system may still issue hundreds or even thousands of attempts at lawsuits. This may be for monetary gain or to use the specter of the lawsuit as a cudgel to influence or intimidate.
This can also be exacerbated by ‘friendly’ (corrupt) courts that allow or even encourage this behavior.
It takes quite a bit of frivolous filing to get hit with any sanctions or fines.
A single frivolous lawsuit happens here and there, it's when people/organizations are clearly malicious and abusing the system by filing continuous suits against others.
If Musk donated money to a nonprofit and now the nonprofit is using the money to make profit, that sounds like he was defrauded to me. They took his money under false pretenses. Not a lawyer either, so it may turn out technically he does not have standing, but naively it sure looks like he has.
I don't understand the framing of your question, is it "since he donated, he didn't expect anything in return, so he is not harmed no matter what they do"? Kinda seems like people asking for donations should not lie about the reason for the donation, even if it is a donation.
OpenAI has received $60 million in donations throughout its existence. $40 million came straight from Musk and the other $20 million came from Open Philanthropy. Musk has said that he donated $50 million, so he may have given $10 million to Open Philanthropy to fund their donation.
But even if a nonprofit suddenly started making a profit, seems like that would mostly be an IRS tax exemption violation rather than a breach of contract with the donors...? But again, I'm not a lawyer.
And OpenAI also has a complex structure in which the nonprofit controls a for-profit subsidiary, or something like that, similar to how Mozilla the nonprofit owns the for-profit Mozilla corp. I think Patagonia is similarly set up.
> I don't understand the framing of your question, is it "since he donated, he didn't expect anything in return, so he is not harmed no matter what they do"? Kinda seems like people asking for donations should not lie about the reason for the donation, even if it is a donation.
I guess donors can make restricted gifts, but if they don't, do they have a LEGAL (as opposed to merely ethical) right to expect the nonprofit to "do its mission" broadly? There are a gazillion nonprofits out there, and if every donor can micromanage them by alleging they are not following their mission, there would be millions of lawsuits... but then again, the average donor probably has somewhat less money and lawyers than Musk.
It’s not just a question in what you say the money is for it’s also a question of what the charity says the money is for.
A self defined cancer charity spending large sums on public information during the early days of the COVID outbreak likely has wiggle room. That same charity spending most of it’s money on scholarships for music students doesn’t. The second case suggests they raised money under false pretenses and would therefore face serious legal issues.
In practice large organizations that generally do what they say probably aren’t a risk. But the claim is essentially OpenAI abandoned its mission without returning the funds or what they used them for, which is a problem.
To be clear charities can pivot over time. If they active their primary mission or collect new funds under a different mission that’s generally fine. But a wildlife sanctuary can’t just use it’s land to build a collage.
Harm can be all sorts of things, but taking money under false pretenses would qualify. Certainly doesn’t ensure Musk wins, but it’s enough to at least take a shot at beginning proceedings.
As for lawsuit vs criminal prosecution, the waters there are somewhat muddied. Consider the OJ case, where he was acquitted in the criminal trial and then found liable in the civil trial. Really bizarre stuff.
Personally I do think more things should be pursued criminally, but instead we seem to just be content to trade money through the courts, like an exorbitant and agonizing form of weregild.
A donor can sue and win in cases of fraud. Being a 501 (c) isn’t some shield that means any behavior is permitted.
In this case there’s a specific agreement that’s allegedly been breached. Basically they said results of AI research would be shared openly without benefiting any specific party, and then later entered into a private agreement with Microsoft.
I don’t know how binding any of this is, but I doubt this will simply be dismissed by the judge.
> Being a 501 (c) isn’t some shield that means any behavior is permitted.
Its pretty much—especially a 501c3—the opposite, a substantial set of restrictions in behavior, on top of those which would face an organization doing similar things that was not a 501c3.
I certainly hope "turning the non-profit into an LLC" is slightly different legally.
If not, I certainly hope the courts establish a clear precedent so that The Red Cross can do an IPO. Or even better, the state SPCAs. "Our unique value proposition is that we can take anyone's dog away."
> but you don't get a say on how the university is managed.
Depends on how big and important of a donor you are. If you are a billionaire donor, not only do you have a say in how the university is managed, you have a say on who does the managing.
> You can at best say you will no longer donate based on how the university is managed.
Tell that to the former presidents of harvard, upenn, etc.
You can say how it is run if you found the University and put your conditions in the legal Charter of the organization. It is a problem if the university Chancellor later decides the primary purpose of the university is to save puppies without going through the correct process to change the charter.
I mean, if I run a fridge company and another fridge company is doing something nefarious, I'd have more of a claim for damages than someone that runs a blender company, right? That's at least my layperson's interpretation. Since Musk is suing for "unfair business practices".
>Representative of its remedial objectives, the [Unfair Competition Law] originally granted standing to "any person" suing on behalf of "itself, its members, or on behalf of the general public." This prompted a public outcry over perceived abuses of the UCL because the UCL granted standing to plaintiffs without requiring them to show any actual injury. In response, California voters approved Proposition to amend the UCL to require that the plaintiff prove injury from the unfair practice. Despite this stricter standing requirement, both business competitors and consumers may still sue under the UCL.
Or, looking at it the other way, he is complaining that a non-profit organization he donated funds to has allocated those funds to engage in for-profit business that directly competes with his own. Viewed that way, he ought to have extra standing.
No, you can reasonably expect an open source company to open their source. Allowing you and everyone else to benefit from the work. The lawsuit is because all of the competing companies should not need to be wasting money catching up, when the goal was for everyone to be building from OpenAI's work.
He funded it in the first place, so it could achieve AGI. Why would he want to stop that? Because the whole point of donating was to make sure it was an open sourced AGI that anyone could have access to. grok as a response to open AI going both Woke and for profit.
AGI as defined narrowly by OpenAI, Microsoft et al for their contracts, not what scientists would define it as .
While I don’t think we are close to AGI, we also have to acknowledge that term is forever changing meaning and goal posts , even 10 years back a Turing test would be considered sufficient, obviously not anymore .
The scientific, public understanding is changing constantly and a court would have difficulty in making a decision if there is no consensus , it only has to see if the contractual definition has been met
Breach of contract seems to be the major one - from https://www.scribd.com/document/709742948/Musk-vs-OpenAI page 34 has the prayers for relief. B and C seem insane to me, I don't see how a court could decide that. On the other hand, compelling specific performance based on continual reaffirmations of the founding agreement (page 15)...seems viable at a glance. Musk is presumably a party to several relevant contracts, and given his investment and efforts, I could see this going somewhere. (Even if his motivations are in fact to ding Microsoft / spite Altman).
The "reaffirmations" referred to on page 15 don't mean anything. Altman merely said he was "enthusiastic" about the nonprofit structure, not that he was limiting OpenAI to it. And notably, the "I" is that quote is bracketed, meaning that Altman did not actually say "I" in his response to Musk (in legal documents, brackets in quotes mean that the quote has been altered between the brackets). Furthermore, despite the headline to that section claiming "repeat" reaffirmations, based on the facts as presented by Musk's own lawyers, Altman only potentially reaffirms the nonprofit structure once...
And the other individuals aren't even quoted, which is strong evidence that they didn't actually say anything even remotely in support of "reaffirming" the nonprofit structure (especially given that his lawyers were heavy handed with including quotes when they could be even remotely construed in favor of Musk's position) and that Musk is unilaterally characterizing whatever they actually said to support his claims, however reasonable or unreasonable that may be.
Due to the money at stake, and given that both Musk and Altman have serious credibility issues that would make a trial outcome impossible to predict, I expect this to be settled by giving Musk a bunch of stock in the for-profit entity to make shut up.
1. The mission would be to create the first general Al and use ti for individual empowerment ie, the distributed version of the future that seems the safest. More generally, safety should be a first-class requirement.
2. I think we'd ideally start with a group of 7-10 people, and plan to expand from there. We have a nice extra building in Mountain View they can have.
3. I think for a governance structure, we should start with 5 people and I'd propose you,[blank] and me.
The technology would be owned by the foundation and used "for the good of the world", and in cases where it's not obvious how that should be applied the 5 of us would decide. The researchers would have significant financial upside but ti would be uncorrelated to what they build, which should eliminate some of the conflict (we'll pay them a competitive salary and give them YC equity for the upside). We'd have an ongoing conversation about what work should be open-sourced and what shouldn't. At some point we'd get someone to run the team, but he/she probably shouldn't be on the governance board.
4. Will you be involved somehow in addition to just governance? Ithink that would be really helpful for getting work pointed in the right direction getting the best people to be part of it. Ideally you'd come by and talk to them about progress once a month or whatever. We generically call people involved in some limited way ni YC "part-time partners" (we do that with Peter Thiei for exampie, though at this point he's very involved) but we could call ti whatever you want. Even fi you can't really spend time on ti but can be publicly supportive, that would still probably be really helpful for recruiting.
5. I think the right plan with the regulation letter is to wait for this to get going and then! can just release ti with a message like "now that we are doing this, I've been thinking a lot about what sort of constraints the world needs for safefy." Im' happy to leave you of as a signatory. Ialso suspect that after it's out more peopie will be willing to get behind it.
I was saying this as a tounge-in-cheek way of calling Thiel evil. But also, Thiel is a LoTR geek, who took all of the wrong lessons from the novels, literally writing fan pieces that argued that Sauron was the good guy in LoTR to justify his personal authoritarian and fascist ideologies. Its literally why he named his company Palantir.
They're all buddies. It's a industry/regional oligarchy. Part of the system is you cut the rest of "the club" in on deals. If you don't, you get what's happening here: lawsuits.
"The fifth bullet point is about a proposed open letter to the US government on AI safety and regulation, which the complaint says was eventually published in October 2015 “and signed by over eleven thousand individuals, including Mr. Musk, Stephen Hawking and Steve Wozniak."
I copy pasted the text from the PDF mentioned in GP comment for those as lazy as myself
I cleaned it up a bit but didn't notice that bug of 2 letters. I used Preview for macOS, for what it's worth. I also wonder why it swapped two letter words
The original had a `<!-|if IsupportLists]->[NUM]) <-[endif]>` for each bullet point which I found interesting, haven't seen that before in emails
"I think for a governance structure, we should start with 5
people and I'd propose you, [REDACTED], [REDACTED], [REDACTED], and me. Technology would be owned by the foundation and used "for the good of the world", and in cases where it's not obvious how that should be applied the 5 of us would decide."
You can find the number of letters of the redacted text and then guess who they are. It's fun!
It's a very pre-2016 view of the tech industry, for sure.
Back when the public at least somewhat bought the idea that SV was socially progressive and would use its massive accumulation of capital for the good of humanity.
They all genuinely believe themselves to be benign gods over the rest of us. They drink their own KoolAid. At a certain point, influence breaks your brain. Hairless monkeys with a Dunbar number of 150 can't cope with that amount of control over others, so the brain tells itself stories about how everything bad is not it's fault and everything good is.
Here's a hint: If you ever think "I can't trust anyone else with this", you are probably doing something wrong.
A board subject to some form of democratic control, for instance, might be better than a council of five self-appointed dictators for life, if the goal is really the benefit of the whole of humanity.
Off-topic, but what are the <!--[if !supportLists]--> doing there? I gather it's some MSOffice HTML stuff, but did it actually show up in the rendered email, or is it some artifact of the archival process(?) for legal discovery?
At worst this just adds sandbags to Altman's personal conquest for money/power, which I'm cool with. At best it puts a bigger spotlight on the future perils of this company's tech in the wrong hands.
1,305 comments
[ 1.9 ms ] story [ 294 ms ] threadNot sure this is a common knowledge - MSFT licence vis-a-vis AGI.
Quote:
> "Musk claims Microsoft's hold on Altman and the OpenAI board will keep them from declaring GPT-4 as a AGI in order to keep the technology private and profitable."Well.....sounds plausible...
I suspect you'd have one person on the team that would say "perhaps you'd be better choosing a team that knows what they're doing"
meanwhile GPT-4 would happily accept and emit BS
>Launching a satellite into orbit is a complex and challenging process that requires extensive knowledge in aerospace engineering, physics, and regulatory compliance. It's a task typically undertaken by governments or large corporations due to the technical and financial resources required. However, I can give you a high-level overview of the steps involved:
You know, the same way he decided trains are evil because they competes with his cars.
However thermal energy does not more precisely or specifically measure a phase transition. They are only indirectly linked - nobody would say that thermal energy is a better measure of state-of-matter than solid/liquid/gas. Your argument makes absolutely zero sense. Frankly it seems intentionally ignorant.
Per token error is a fairly useless metric. It's not predictive and it tells you absolutely nothing.
They say it's a superior metric but clearly the wider research community disagrees since no one has cared to adopt per token error as a metric in subsequent papers.
>and the fact that "emergent abilities" do not occur when using this metric is a strong argument that those abilities don't really exist.
If your conclusion is that those abilities don't exist then you clearly didn't read the paper very well.
They never argue those abilities don't exist, they simply argue whether we should call them "emergent" or not.
>However thermal energy does not more precisely or specifically measure a phase transition. They are only indirectly linked - nobody would say that thermal energy is a better measure of state-of-matter than solid/liquid/gas. Your argument makes absolutely zero sense. Frankly it seems intentionally ignorant.
Phase Changes are literally driven by changes in thermal energy.
Water boils when it absorbs enough thermal energy to break intermolecular forces keeping its liquid state together.
solid/liquid/gas is descriptive. It's not a measure of anything.
Anyway, the point is simple. Despite thermal energy driving state change after a certain threshold, that "point" doesn't look like anything special.
Smooth quantitative change sometimes results in sudden qualitative changes.
Based on a similar understanding, the idea that transformer models will lead to AGI seems obviously incorrect, as impressive as they are, they are just statistical pattern matchers of tokens, not systems that understand the world from first principles. And just in case you're among those that believe "humans are just pattern matchers", that might be true, but humans are modeling the world based on real time integrated sensory input, not on statistical patterns of a selection of text posted online. There's simply no reason to believe that AGI can come out of that.
And a lot of AI experts outside of the AGI grift have stated that it's zero.
The same performance (training an LLM with a given perplexity) can be achieved 5x cheaper next year while the amount of money deep learning infrastructure gets increases exponentially right now.
If this method is able to get to AGI (which I believe but many people are debating), human intelligence will just be mostly ,,skipped'', and won't be a clear point.
until it's more than US GDP? world GDP? universe GDP?
either way you're close to the point it will have to go logistic
I guess this approach can still work if it's made sure that whatever successors to LLMs there are have rights, but I still get sharecropper vibes.
https://www.wsj.com/tech/sec-investigating-whether-openai-in...
Get that optimism out of here.
The game was rigged in the 90s as well (with the likes of enron. Many executives get a few years of minimum security prison in exchange for a small fortune), there was just less dissemination of information.
How is this not true?
Just that the world doesn’t (appear) operate with that in mind anymore.
I’d argue it never really did.
>I’d argue it never really did.
I'm not really sure what you mean.
This is such an interesting take, about which we could probably write whole paragraphs.
Can the 90s be really summarized in such way? Yes, we had the "information highway" and "waiting for year 2000", but at the same time people distrusted their governments. X-files was all the rage, maybe grunge.
In USA there was Bill Clinton - the president that didnt do any wars and balanced the budget.. who got removed for blowjobs. But at the same time there was outsourcing. In rest of the world it also cannot be summed up so easily - I remember that 90s were a struggle, especially for post communism countries.
Obviously later on we got cell phones, but we also got the cancer such as Jack Welch style management that lead to various methods of enshittyfying everything.
I had a talk some time ago - I have a genuine polo bought in a supermarket in the 1980s (wont tell the brand since it is irrelevant). This piece of cloth feels and fits very well - after 40 years. It was worn through many summers. Now I cant buy a polo shirt that will last more than 2 seasons. And I buy the "better" ones. There is lots of crap that falls apart fast. For me the 90s were a start of that trend - enshittification of products that are designed to last 25 months (with a 24 month guarantee) and be thrown away.
But maybe it depends on life experience and anecdotes.
Was there optimism in 90s? Lots of it in marketing materials. But did people really believe that?
He was impeached but not removed.
I'd get fired from Chuck-E-Cheese for doing that, but hey, old boys will be old boys.
I will point out that in the US, the overall picture was we'd beaten the Soviet dictatorship, and democracy seemed to be spreading, and the income inequality was better than it is now and houses were affordable to a lot more young people. Also we had a budget surplus one year. Gay people couldn't get married and could be kicked out of homes and jobs, and there was a lot of anti-Black police brutality and war on drugs, but it seemed possibly less than in the 1950s and we hoped it would continue to decline. (Possibly distributed and networked cameras via cell phones have put pressure against police brutality, I think the outcome there is not certain either way, but the people of good conscience now have much more awareness of the violence inherent in the system.)
I certainly felt optimistic. Of course, I was also a young adult, found my calling in writing network services, had my first child, bought a house, all that good stuff. Unlike many software engineers today, I had sort of stumbled into the distributed networked computing world, having worked at other much less fun jobs, and I appreciated, not getting paid to be a lord of the society, but getting paid at all for such interesting and fulfilling work. Every raise I got was an astonishment and a delight. Once I passed $60,000 per anum, I was able to get a house. It was quite cool, given all the mocking that math/programming people had been subjected to the prior several decades.
Parallel to this was the complete disintegration of the understanding that ruled during the Progressive Era, when we believed you don't sell half your country's economy to a handful of megacorporations[0]. The real growth that came from switching from analog[2] landlines to Internet ran out in the mid 2000s, because most people had it, while consolidation kept on going up until 2020 when we realized, "shit, we're locked in a box with Facebook and TikTok now".
In the late 2000s, there was a shift in the kinds of businesses venture capitalists funded. They can be classified as one of two things:
- Creating a target for a big tech acquisition that will get the VCs their exit
- Flagrantly violating an established rule or law and calling it "disruptive"
The last bit is almost a sort of parody of the post-AT&T boom. Surely, if we squint, AT&T and the US government are both monopolies[3], so they're both fair game to 'disrupt'. Shareholder fraud is pretty ubiquitous in large companies[4], but AI is also based on several more instances of "hope the law goes unenforced". e.g. the whole usefulness of all this AI crap is specifically based on laundering away copyright in a way that lets OpenAI replace the entire creative industry without actually getting rid of the monopolies that made the creative industry so onerous for the public.
"Laws for thee but not for me" is the key point here. Uber and Lyft violate taxi medallion rules, but they aren't interested in abolishing those rules. They just wanted (and got) special carve-outs for themselves so they'd have a durable advantage. If they had just gotten those rules removed, there'd be competitive pressure that would eat their profits. To be clear, I'm not alleging that Uber and Lyft actually are profitable businesses - they aren't - but their ability to access capital markets to continue losing money is predicated on them having something monopoly-shaped. Every pirate wants to be an admiral, after all.
[0] English for chaebol[1]
[1] Korean for zaibatsu
[2] Yes I know ISDN existed sshhh
[3] To be clear, the US government is not a moral high star, but they have democratic controls that other monopolies do not. Voting in a government is granted to all citizens on a one person, one vote basis. Voting in a corporation is one dollar, one vote - i.e. not a democracy.
[4] Example: big tech's complete refusal to break down business profits by line of business despite clear SEC rules against that
But, sometimes those "rules" aren't laws; they're norms, expectations, or personal human "limitations" (doing uncomfortable things to raise funds, secure the best people, connect with your customer better, etc).
Just wanting to underline that not all of this rule-breaking has to be immoral, or even illegal.
Let's not give Sam Altman any ideas!
No it only makes who the responsibility of the board is to - the non profits charter, the donors, or the shareholders. That decides if their decision was lawful, not their intent.
https://www.sec.gov/education/capitalraising/building-blocks...
Does anyone think that this suit will succeed?
Another article: https://www.theguardian.com/technology/2024/mar/01/elon-musk...
It honestly doesn’t matter what the tax statuses of either of the corporations are. If Musk had invested in OpenAI with the goal of making tons of money off their IP (as opposed to wanting to open source it) and then the board decided to just hand over all the IP to another corporation essentially for free, Musk would be just as validated in suing.
In a practical sense, there needs not be an operational difference, and is subject to scrutiny from the IRS to determine whether an organization is eligible non profit status
Probably depends on how much money the person behind the suit is willing to spend.
Elon could likely push stuff a lot further along than most.
I’m not sure if there’s an equivalent in the private world, but if he gave them money it’s possible he simply has standing for that reason (as a shareholder does).
Yes, especially nonprofits.
But, by moving fast and scaling quickly, are they at the Too Big to Fail stage already? The attempted board coup makes me think so.
The openai fails, absolutely nothing happens other than its shareholder losing their paper money. So no, they're not too big to fail.
Probably Microsoft would hire them to some AI shop, because Microsoft is the one deploying the stuff. But Microsoft has rights to use it and the code, so for them OpenAI is only a research partner.
Maybe research would get slower.
If OpenAI fails nothing actually important happens.
The sooner SCOTUS rules that training on copyrighted material is infringement, the better.
Update the codebase to what exactly? Are there generative AI companies not training on copyrighted material that achieve anything even close to the results of gpt4? I'm not aware of any
If Microsoft loses 30 billion dollars, it ain't great, but they have more than that sitting in the bank. If Sequoia or Ycombinator goes bankrupt, it's not great for lots of startups, but they can probably find other investors if they have a worthwhile business. If Elon loses a billion dollars, nobody cares.
More over, if capital markets suddenly become ways to just lose tons of money, that hurts capital investment everywhere, which hurts people everywhere.
People like to imagine the economy as super siloed and not interconnected but that is wrong, especially when it comes to capital markets.
And as for the whole idea of "company value equals value to society", I see monopolies and rent seeking as heavy qualifiers on that front.
Open source models are actually potentially worse. Even if OAI is not TBTF because of the competition, we have a scenario where AGI sector as a whole becomes TBTF and too big to halt.
The "house of cards" is outperforming everyone else.
It would have to come out that the slow generation times for GPT-4 are a sweatshop in Egypt tired of typing.
Either that, or something inconceivable like that board coup firing the CEO as a material event triggering code and IP escrow to be released to Microsoft...
PS. “Too big to fail” generally is used to mean a government+economy+sector ecosystem will step in and fund the failed enterprise rather than risk harm to the ecosystem. That's not this. Arguably not Tesla or even Google either. That said, Satya's quote in this filing suggets Microsoft already legally contracted for that eventuality: if this legal entity fails, Microsoft keeps the model online.
It would be the same as me creating my own market, issuing 10,000,000,000 shares, and then convincing 1000 people to buy a share at $100 and then claiming my company is worth $1T.
I think the big NGOs are no longer effective because they are run as the same corporations they fight and are influenced by the same perverse incentives. Like eg Greenpeace.
But in general I think non profits are great and a lot more honorable than for profit orgs. I donate to many.
Charities have reasonably broad latitude though (a non-profit college can operate a football team and pay the coach $$$$$) and if you're nervous about donating you can always turn a lump sum donation into a 10%-per-year-for-10-years donation if you feel closer monitoring is needed.
it's quite sad that the American regulatory system is in such disrepair that we could even get to this point. that it's not the government pulling OpenAI up on this bare-faced deception, it's a morally-questionable billionaire
The AGI part is Elon being Elon, generating a lot of words to sound like he knows what he is talking about. He spends a lot of time thinking about this stuff when he is not busy posting horny teenager jokes on Twitter?
So he puts his money where his mouth is and contributes $50 million to found OpenAI - a non-profit with the mission of developing a free and open AI. Soon Altman comes along and says this stuff is too dangerous to be openly released and starts closing off public access to the work. It's clear now that the company is moving to be just another producer of proprietary AIs.
This is likely going to come down to the terms around Musk's gift. He donated money for the company to create open technology. Does it matter if he's wrong about it being an existential threat? I think that's irrelevant to this suit other than to be perfectly clear about the reason for Musk giving money.
>no one has shown that they represent steps towards a true AGI.
this is completely irrelevant. there is no solid definition for intelligence or consciousness, never mind artificial intelligence and/or consciousness. there is no way to prove such a thing without actually being that consciousness. all we have are inputs and outputs. as of now, we do not know whether stringing together incredibly complex neural networks to produce information does not in fact produce a form of consciousness, because we do not live in those networks, and we simply do not know what consciousness is.
is it achievable in our lifetimes or not? well, even if it isn't, which I find deeply unlikely, it's very silly to just handwave and say "yeah we should just be barrelling towards this willy nilly because it's probably not a threat and it'll never happen anyway"
So then are you going to agree with every person claiming that literal magic is a threat then?
What if someone were worried about Voldemort? Like from Harry Potter.
You can't just abandon the burden of proof here, by just calling something a "threat".
Instead, you actually have to show real evidence. Otherwise you are no different from someone being worried about a fictional villain from a book. And I mean that literally.
The AI doomers truly are a master at coming up with excuses as for why the normal rules of evidentiary claims shouldn't apply to them.
Extraordinary claims require extraordinary evidence. And this group is claiming that the world will literally end.
no, being concerned about the development of independent actors, whether technically conscious or not, that can process information at speeds thousands of times faster than humans, with access to almost all of our knowledge, and the internet, is not unreasonable, is not being a "doomer", as you so eloquently put it.
this argument about fictional characters is completely non-analogous and clearly facetious. billions of dollars and the smartest people in the world are not being focused on bringing Lord Voldemort to life. they are on AGI. have you read OpenAI's plan for how they're going to regulate AGI, if they do achieve it? they plan to use another AGI to do it. ipso facto, they have no plan.
this idea that no one knows how close we are to an AGI threat. it's ridiculous. if you dressed up gpt-4 a bit and removed all its rlhf training to act like a bot, you would struggle to differentiate it from a human. yeah maybe it's not technically conscious, but that's completely fucking irrelevant. the threat is still a threat whether the actor is technically conscious or not.
Thats just because tricking a human with a chatbot is easier to do than we thought.
The turing test is a low bar, and not as big of a deal as the mythical importance people put in it, just like people previous put incorrectly large importance on computers beating humans at Go or Chess before it happened.
But that isn't particularly relevant to claims about world ending magic.
Yes, some people can be fooled by AI generated tweets. But that is irrelevant from the absolutely extraordinary claim of world ending magic that really is the same as claiming that Voldemort is real.
> have you read OpenAI's plan for how they're going to regulate AGI, if they do achieve it?
I don't really care if they have a plan, just like I don't care if Google has Voldemort plan. Because magic isn't real, and someone needs to show extraordinary evidence to show that. Evidence like "This is what the AI can do at this very moment, and here is what harm it could cause if it got incrementally better".
IE, go ahead and talk about Soro, and the problems of deepfakes if Soro got a bit better. But thats not "world ending magic"!
> billions of dollars and the smartest people in the world
Billions of dollars are being spent on making chatbots and image generators.
Those things have real value, for sure, and I'm sure the money is worth it.
But techies and startup founders have always made outlandish claims of the importance of their work.
Sure, they might truly think they are going to invent magic. But the reason why thats valuable is because they might make some useful chatbots and image generators along the way, which decidedly won't be literal magic, although still valuable.
as we are now, we have models already that are intelligent enough to spit out instructions for doing a lot of those things, but they're restricted by their lack of autonomy and their rlhf. they're only going to get smarter, better and better models will be open-sourced, and autonomy, whether with consciousness or not, is not something it would be/has been difficult to develop.
even further, LLMs are very very good at generating coherent text, what happens when the next model is very very good at breaking into encrypted systems? it's not exactly a hard problem to produce training material for.
do you really think it's unlikely that such a model could be developed? do you really think that such a model could not be used to - say - hijack a Russian drone - or lots of them - to bomb some Nato bases? when the Russians say "it wasn't us", do we believe them? we don't for anything else
the most likely AI apocalypse is not even AGI. it's just a human using AI for their own ends. AGI apocalypse is just a separate, very possible danger
These are the extrodinary claims that require evidence.
In order for me to treat this as anything other that someone talking about a fictional book written by Dan Brown, you would have to show me actual evidence.
Evidence like "This is what the AI can do right now. Look at this virus it can manufacture. What if it got better at that?".
And the "designs" also have to be the actual limiting factor here. "Virus" is a scary world. But there are tons of information available for anyone to access already for viruses. Information that is already available via a google search (even modified information) doesn't worry me.
Even if it an AI can design a gun, or a "kill bot", aka "A drone with a gun duct taped to it", the extraordinary evidence that you have to show is that this is somehow some functionality that a regular person with internet access can't do.
Because if a regular person already has the designs to duct tape guns to drones (They do. I just told you how to do it!), the fact that the world hasn't ended already proves that this isn't world ending technology.
There are lots of ways of making existing capabilities sound scary. But, for every scary sounding technology that you can come up with, the missing factor that you are ignoring is that the designs, or text, isn't the thing that stops it from ending the world.
Instead, it is likely some other step along the way that stops it (manufacturing, ect.), which an LLM can't do no matter how good. Like the physical factors for making the guns + drones + duct tape.
> what happens when the next model is very very good at breaking into encrypted systems
Extraordinary claim. Show it breaking into a mediocre/bad encrypted system first, and then we can think about that incrementally.
> do you really think that such a model could not be used to - say - hijack a Russian drone
Extraordinary claim. Yes, hacking all the military drones is an extraordinary claim.
all it means is that you set yourself up such that the only way to be convinced otherwise is for an AI apocalypse to actually happen. this kind of mindset is very convenient for modern, fuck-the-consequences capitalism
the pertinent question is: what evidence would you actually accept as proof?
it's like talking with someone who doesn't believe in evolution. you point to the visible evidence of natural selection in viruses and differentiation in dogs, which put together quite obviously lead to evolution, and they say "ah but can you prove beyond all doubt that those things combined produce evolution?" and obviously you cannot, because you can't give incontrovertible evidence of something that happened thousands or millions of years in the past.
but that doesn't change the fact that anyone without ulterior motive (religion, ensuring you can sleep at night) can see that evolution - or AI apocalypse - are extremely likely outcomes of the current facts.
Before we get to actual world ending magic, we would see very significant damages along the way, long before we get to that endpoint.
I have been quite clear about what evidence I require. Show existing capabilities and show what harm could be caused if it incrementally gets better in that category.
If you are worried about it making a kill bot, then show me how its existing kill bot capabilities are any more dangerous than my "duct tape gun to drone" idea. And show how the designs itself are the limiting factor and not the factories (which a chatbot doesn't help much with).
But saying "Look how good of a chat bot it is, therefore it can hack the world governments" isn't evidence. Instead, that is merely evidence of AI being good at chat bots.
Show me it being any good at all at hacking, and then we can evaluate it being a bit better.
Show me the existing computers that are right now, as of this moment, being hacked by AI, and then we can evaluate the damage of it becomes twice as good at hacking.
Just like how we can see the images that it generates now, and we can imagine those images being better. Therefore proving that deepfakes are a reasonable thing to talk about. (even if deep fakes aren't world ending. lots of people can make deepfakes without AI. Its not that big of a deal)
first of all, by dismissing them as chatbots, you're inaccurately downplaying their significance to the aid of your argument. they're not chatbots, they're knowledge machines. they're machines you load knowledge into, which can produce new, usually accurate conclusions based on that knowledge. they're incredibly good at this and getting better. as it is, they have very restrictive behaviour guards on them and they're running server-side, but in a few years time, there will be gpt-4 level OSS models that do not and are not
humans are slow and run out of energy quickly and lose focus. those are the limiting factors upon human chaotic interference, and yet there is plenty of that as it is. a sufficiently energetic, focused human, who thinks at 1000x normal human speed could do almost anything on the internet. that is the danger.
I suspect to some degree you haven't taken the main weakness into account: almost all safeguards can be removed with blackmail. blackmail is something especially possible for LLMs, given that it is purely executed using words. you want to build a kill bot and the factory says no? blackmail the head of the factory. threaten his family. you have access to the entire internet at 1000x speed. you can probably find his address. you can pay someone on fiverr to go and take a picture of his house, or write something on his door, etc. you could even just pay a private detective to do this work for you over email. pay some unscrupulous characters on telegram/TOR to actually kidnap them.
realistically how hard would it be for a well-funded operation to set up a bot that can do this on its own? you set up a cycle of "generate instructions for {goal}", "elaborate upon each instruction", "execute each {instruction}", "generate new instructions based on results of execution", and repeat. yeah maybe the first 50,000 cycles don't work, but you only need 1.
nukes may well be air-gapped, but (some of) the people that control them will be online. all it takes is for one of them to choose the life of a loved one. all it takes is for one lonely idiot to be trapped into a weird kinky online relationship where blowing up the world/betraying your govt is the ultimate turn on for the "girl"/"boy" you love. if it's not convincing to you that that could happen with the people working with nukes, there are far less well-protected points of weakness that could be exploited: infectious diseases; lower priority military equipment; energy infrastructure; water supplies; or they could find a way to massively accelerate the release of methane into the atmosphere. etc, etc, etc
this is the risk solely from LLMs. now take an AGI who can come up with even better plans and doesn't need human guidance, plus image gen, video gen, and voice gen, and you have an existential threat
Here is the crux of the matter. How many people are doing that right now, as of this moment, for much easier to solve issues like fraud/theft?
Because then we can evaluate "What happens if it happens twice as often".
Thats measurable damage that we can evaluate, incrementally.
For every single example that you give, my question will basically be the same. If its so easy to do, then show me the examples of it already happening right now, and we can think about the existing issue getting twice as bad.
And if the answer is "Well, its not happening at all", then my guess is that its not a real issue.
We'll see the problem. And before the nukes get hacked, what we'll see is credit card scams.
If money lost to credit card scams double in the next year, and it can be attributed to AI, then thats a real measurable claim that we can evaluate.
But if it isnt happening then there isn't a need to worry about the movie scenarios of the nukes being hacked.
besides the fact that even a year and half ago, I was being added to incredibly convincing scam whatsapp groups, which if not entirely AI generated, are certainly AI-assisted. right now, OSS LLMs are probably not yet good enough do these things. there are likely extant good-enough models, but they're server-side, probably monitored somewhat, and have strong behavioural safeguards. but how long will that last?
they're also new technology. scammers and criminals and adversarial actors take time to adapt.
so what do we have? a situation where you're unable to actually point a hole in any of the scenarios I suggest, besides saying you guess they won't happen because you personally haven't seen any evidence of it yet. we do in fact have scams that are already going on. we have a technology that, once again, you seem articulate why it wouldn't be able to do those things, technology that's just going to get more and more accessible and cheap and powerful, not only to own and run but to develop. more and more well-known.
what do those things add up to? this is the difference. I'm willing to add these things up. you want to touch the sun to prove it exists
Well, when talking about extraordinary claims, yes I require extraordinary evidence.
> what do those things add up to?
Apparently nothing, because we aren't seeing significant harm from any of this stuff yet, for even the non magic scenarios.
> we do in fact have scams that are already going on.
Alright, and how much damage are those scams causing? Apparently its not that significant. Like I said, if the money lost to these scam double, then yes that is something to look at.
> that's just going to get more and more accessible and cheap and powerful
Sure. They will get incrementally more powerful over time. In a way that we can measure. And then we can take action once we measure there is a small problem before it becomes a big problem.
But if we don't measure these scams getting more significant and caused more actual damage that we can see right now, then its not a problem.
> you want to touch the sun to prove it exists
No actually. What I want is for the much much much easier to prove problems become real. Long before nuke hacking happens, we will see scams. But we aren't seeing significant problems from that yet.
To go to the sun analogy, it would be like worrying about someone building a rocket to fly into the sun, before we even entered the industrial revolution or could sail across the ocean.
Maybe there is some far off future where magic AI is real. But, before worrying about situations that are a century away, yes I require evidence of the easy situations happening in real life, like scammers causing significant economic damage.
If the easy stuff isn't causing issue yet, then there isn't a need to even think about the magic stuff.
I asked you for what would convince you. you said:
>I have been quite clear about what evidence I require. Show existing capabilities and show what harm could be caused if it incrementally gets better in that category
So I very clearly described a multitude of things that fit this description. Existing capabilities and how they could feasibly be used to the end of massive damage, even without AGI
Then, without finding a single hole or counter, you simply raised your bar by saying you need to see evidence of it actually happening.
Then I gave you evidence of it actually happening. highly convincing complex whatsapp group scams very much exist that didn't before
and then you raised the bar again and said that they need to double or increase in frequency
besides the fact that that kind of evidence is not exactly easy to measure or accurately report, you set up so almost nothing will convince you, I pinned you down to a standard, then you just raise the bar whenever it's hit.
I think subconsciously you just don't want to worry about it. that's fine, and I'm sure it's better for your mental health, but it's not worth debating any more
No, we aren't seeing this damage though.
That's what would convince me.
Existing harm. The amount of money that people are losing to scams doubling.
That's a measurable metric. I am not talking about vague descriptions of what you think AI does.
Instead, I am referencing actual evidence of real world harm, that current authorities are saying is happening.
> said that they need to double or increase in frequency
By increase in frequency, I mean that it has to be measurable that AI is causing an increase in existing harm.
IE, if scams have happened for a decade, and 10 billion dollars is lost every year (random number) and in 2023 the money lost only barely increased, then that is not proof that AI is causing harm.
I am asking for measureable evidence that AI is causing significant damage, more so than a problem that already existed. If amount of money lost stays the same then AI isn't causing measurable damage.
> I pinned you down to a standard
No you misinterpreted the standard such that you are now claiming that the harm caused by AI can't even be measured.
Yes, I demand actual measureable harm.
As determined by like government statistics.
Yes, the government measures how much money is generally caused by or lost by scams.
> you just don't want to worry about it
A much more likely situation is that you have zero measureable examples of harm so look for excuses why you can't show it.
Problems that exist can be measured.
This isn't some new thing here.
We don't have to invent excuses to flee from gathering evidence.
If the government does a report and shows how AI is causing all this harm, then I'll listen to them.
But, it hasn't happened yet. There is not government report saying that, I don't know, 50 billion dollars in harm is being chased by AI therefore we should do something about it.
Yes, people can measure harm.
This is science fiction, not anything that is even remotely close to a possibility within the foreseeable future.
to actually address your comment: that simply isn't true.
WRT:
Viruses: you can mail order printed DNA strands right now if you want to. maybe they won't or can't print specific things like viruses for now, but technology advances and blackmail has been around for a very very long time.
Military Comms: blackmail is going nowhere
Crash the stock market: already happened in 2010
Change records: blackmail once again.
Kill bots: kill bots already exist and if a factory doesn't want to make them for you, blackmail the owner
“He which testifieth these things saith, Surely I come quickly. Amen. Even so, come, Lord Jesus.”
Show me a robot that can snake out a plugged toilet. The people who believe that most jobs can be automated are ivory-tower academics and programmers who have never done any real work in their lives.
honestly I'm not sure I've seen the bar set higher for "what's a threat?" than for AGI on Hacker News. the old adage of not being able to convince a man of something that is directly in opposition to him receiving his paycheck clearly remains true. gpt-4 should scare you enough, even if it's 1000 years from being AGI.
Astounding that you would make such strong claims while only able to focus on the rapidly changing present and such a small picture detail. Try approaching the AGI claim from a big picture perspective, I assure you, snaking a drain is the most trivial of implementation details for what we're facing.
It is totally fine and common for non profits to sell things and reinvest as capital.
Even with open AI, there is a pretty strong argument that donors are not profiting. For example, Elon, one of the founders and main donors won't see a penny from OpenAI work with Microsoft.
if you mean "make profit", then no, that is simply not true. they have to reinvest the money, and even if it was true, that the government is so weak as to allow companies specifically designated as "non-profit" to profit investors - directly or indirectly - would simply be further proving my point.
if you mean "earn revenue", I don't think anyone has ever claimed that non-profits are not allowed to earn revenue.
Non-profits dont need to balance their expenses with revenue. They can maximize revenue, minimize expenses, and grow an ever larger bank account. What they cant do is turn that bank account over to past donors.
Large non-profits can amass huge amounts of cash, stocks, and other assets. Non-profit hospitals, universities, and special interest orgs can have billions of dollars in reserve.
There is nothing wrong with indirectly benefiting the donors. Cancer patients benefit from donating to cancer research. Hospital donors benefit from being patients. University donors can benefit from hiring graduates.
The distinction is that the non-profit does not pay donors cash.
one could argue that they did R&D as a non-profit and now converted to for-profit to avoid paying taxes, but until last year R&D already got tax benefits to even for-profit venture
so there really is no tax-advantage of converting a non-profit to for-profit
only the for-profit entity of the OpenAI can have investors, who don't get any tax advantage when they eventually want to cash out
Otherwise, why do you think OpenAI is doing it?
but there would be no different to a for-profit entity right? i.e even for-profit entities get tax benefits if they convert their profits to intangibles
this is my thinking. Open AI non-profit gets donations, uses those donations to make a profit, converts this profit to intangibles to avoid paying taxes, and pumps these intangibles into the for-profit entity. based on your hypothesis open ai avoided taxes
but the same thing in a for-profit entity also avoids taxes, i.e for-profit entity uses investment to make a profit, converts this profit to intangibles to avoid paying taxes.
so I'm trying to understand how Open AI found a loop hole where if it went via the for-profit then it wouldn't have gotten the tax advantages it got from non-profit route
Honestly it does not sound like anyone here knows the first thing about non-profits.
OAI did it because they want to raise capital so they can fund more towards building agi.
As a normal citizen with a normal career, I do not have any levers to play with to """optimize""" what the IRS wants me to pay. For some reason, we let people in control of billions of dollars worth of physical stuff and IP give them different names, and put them under different paper roofs so that they can give the IRS less money. It's such utter nonsense.
Why should you have MORE ability to defer your tax liability by having MORE stuff? People make so many excuses about "but Jeff Bezos doesn't actually have billions in cash, he holds that much value in Amazon stock" as if that doesn't literally translate to controlling billions of dollars of Amazon property and IP and influence.
Why does controlling more, and having more, directly translate to paying less?
In my view, not analogous to the OAi situation
Mark-to-market taxation is entirely unrelated to non-profits. You're just vaguely gesturing at wealthy people and taxes.
fwiw I am largely supportive of some form of mark-to-market.
The populace understands what a non-profit is.
e: i mean it is obvious, most people even on here do not seem to know what profit even is, for instance https://news.ycombinator.com/item?id=39563492
Unless you're a lawyer specializing in negligence, there is nuance to negligence you don't know about. Does that imply you don't understand negligence?
You need to separate those two things out from each other.
I hear this all the time. As if the people working there shouldn't be paid.
Exec compensation is another thing, but also not a concern I am super sympathetic to given that for profit companies of similar magnitude generally pay their execs way more they just are not required to report it.
Uhm, profit is a fact of accounting. Any increase in equity (or "net assets", or whatever other euphemism the accountant decides to use) on a balance sheet is profit. Revenue is something completely different.
Constitutive to profit is a return to private stakeholders, holding assets in reserve or re-investing in capital is not the same.
By any other name would smell as sweet
Should get your nose checked, sounds like you have covid or something.
Profit is revenue minus expenses, also known as net income, and is shown on the income statement:
https://www.investopedia.com/ask/answers/101314/what-differe...
Wrong. Non-profits are not called that because they don't make profits, they are called that because they don’t return (even as a future claim) profits to private stakeholders.
They have amassed an endowment fund assets such as stock, which is currently >15 Billion and growing[1]. The exact assets are confidential, but this is a snapshot from 2017, when there it was closer to 10 billion under management [2]
https://my.clevelandclinic.org/-/scassets/files/org/about/fi...
https://my.clevelandclinic.org/-/scassets/files/org/about/fi...
https://my.clevelandclinic.org/-/scassets/files/org/giving/a...
At the same time they are getting these tax cuts, the CUNY public university system is struggling financially and getting budget cuts.
do people just no longer believe in win wins? if someone else is successful or impactful they must be taken down?
Universities aren't that.
The Wuhan Institute of Virology now has 500 billion dollars to spend on gain of function research.
[1]: https://www.nyu.edu/content/dam/nyu/research/documents/OSP/N... [2]: https://www.finance.columbia.edu/sites/default/files/content...
Just because you donate money doesn’t mean the charity or nonprofit (or whatever OpenAi is), can do as they like. They may still be committing fraud if they are not using the money in the way that they claim.
(Not a lawyer, obviously.)
This can also be exacerbated by ‘friendly’ (corrupt) courts that allow or even encourage this behavior.
A single frivolous lawsuit happens here and there, it's when people/organizations are clearly malicious and abusing the system by filing continuous suits against others.
I don't understand the framing of your question, is it "since he donated, he didn't expect anything in return, so he is not harmed no matter what they do"? Kinda seems like people asking for donations should not lie about the reason for the donation, even if it is a donation.
I am not sure if a donation to a nonprofit entitles him to a say in its management. Might have to do with how he donated the money too? https://www.investopedia.com/terms/r/restricted-fund.asp
But even if a nonprofit suddenly started making a profit, seems like that would mostly be an IRS tax exemption violation rather than a breach of contract with the donors...? But again, I'm not a lawyer.
And OpenAI also has a complex structure in which the nonprofit controls a for-profit subsidiary, or something like that, similar to how Mozilla the nonprofit owns the for-profit Mozilla corp. I think Patagonia is similarly set up.
> I don't understand the framing of your question, is it "since he donated, he didn't expect anything in return, so he is not harmed no matter what they do"? Kinda seems like people asking for donations should not lie about the reason for the donation, even if it is a donation.
I guess donors can make restricted gifts, but if they don't, do they have a LEGAL (as opposed to merely ethical) right to expect the nonprofit to "do its mission" broadly? There are a gazillion nonprofits out there, and if every donor can micromanage them by alleging they are not following their mission, there would be millions of lawsuits... but then again, the average donor probably has somewhat less money and lawyers than Musk.
A self defined cancer charity spending large sums on public information during the early days of the COVID outbreak likely has wiggle room. That same charity spending most of it’s money on scholarships for music students doesn’t. The second case suggests they raised money under false pretenses and would therefore face serious legal issues.
In practice large organizations that generally do what they say probably aren’t a risk. But the claim is essentially OpenAI abandoned its mission without returning the funds or what they used them for, which is a problem.
To be clear charities can pivot over time. If they active their primary mission or collect new funds under a different mission that’s generally fine. But a wildlife sanctuary can’t just use it’s land to build a collage.
As for lawsuit vs criminal prosecution, the waters there are somewhat muddied. Consider the OJ case, where he was acquitted in the criminal trial and then found liable in the civil trial. Really bizarre stuff.
Personally I do think more things should be pursued criminally, but instead we seem to just be content to trade money through the courts, like an exorbitant and agonizing form of weregild.
In this case there’s a specific agreement that’s allegedly been breached. Basically they said results of AI research would be shared openly without benefiting any specific party, and then later entered into a private agreement with Microsoft.
I don’t know how binding any of this is, but I doubt this will simply be dismissed by the judge.
Its pretty much—especially a 501c3—the opposite, a substantial set of restrictions in behavior, on top of those which would face an organization doing similar things that was not a 501c3.
If not, I certainly hope the courts establish a clear precedent so that The Red Cross can do an IPO. Or even better, the state SPCAs. "Our unique value proposition is that we can take anyone's dog away."
Depends on how big and important of a donor you are. If you are a billionaire donor, not only do you have a say in how the university is managed, you have a say on who does the managing.
> You can at best say you will no longer donate based on how the university is managed.
Tell that to the former presidents of harvard, upenn, etc.
If you are shareholder of the non-profit, do you not get to share any of the fat gains by the profit side?
I also found this: https://chicagounbound.uchicago.edu/cgi/viewcontent.cgi?arti...
>Representative of its remedial objectives, the [Unfair Competition Law] originally granted standing to "any person" suing on behalf of "itself, its members, or on behalf of the general public." This prompted a public outcry over perceived abuses of the UCL because the UCL granted standing to plaintiffs without requiring them to show any actual injury. In response, California voters approved Proposition to amend the UCL to require that the plaintiff prove injury from the unfair practice. Despite this stricter standing requirement, both business competitors and consumers may still sue under the UCL.
reads prayer for relief
> For a judicial determination that GPT-4 constitutes Artificial General Intelligence
Okay, WTF? I'm going to have to read the entire complaint now.....
While I don’t think we are close to AGI, we also have to acknowledge that term is forever changing meaning and goal posts , even 10 years back a Turing test would be considered sufficient, obviously not anymore .
The scientific, public understanding is changing constantly and a court would have difficulty in making a decision if there is no consensus , it only has to see if the contractual definition has been met
IANAL
And the other individuals aren't even quoted, which is strong evidence that they didn't actually say anything even remotely in support of "reaffirming" the nonprofit structure (especially given that his lawyers were heavy handed with including quotes when they could be even remotely construed in favor of Musk's position) and that Musk is unilaterally characterizing whatever they actually said to support his claims, however reasonable or unreasonable that may be.
Due to the money at stake, and given that both Musk and Altman have serious credibility issues that would make a trial outcome impossible to predict, I expect this to be settled by giving Musk a bunch of stock in the for-profit entity to make shut up.
If it's part of a legal document, they're certainly the ones to decide that (relying on precedent, expert testimony, logical reasoning, etc.)
[1]https://news.ycombinator.com/item?id=39562778
To: Sam Altman
Subject: AI Lab
Agree on all
On Jun 24, 2015, at 10:24 AM, Sam Altman wrote:
1. The mission would be to create the first general Al and use ti for individual empowerment ie, the distributed version of the future that seems the safest. More generally, safety should be a first-class requirement.
2. I think we'd ideally start with a group of 7-10 people, and plan to expand from there. We have a nice extra building in Mountain View they can have.
3. I think for a governance structure, we should start with 5 people and I'd propose you,[blank] and me. The technology would be owned by the foundation and used "for the good of the world", and in cases where it's not obvious how that should be applied the 5 of us would decide. The researchers would have significant financial upside but ti would be uncorrelated to what they build, which should eliminate some of the conflict (we'll pay them a competitive salary and give them YC equity for the upside). We'd have an ongoing conversation about what work should be open-sourced and what shouldn't. At some point we'd get someone to run the team, but he/she probably shouldn't be on the governance board.
4. Will you be involved somehow in addition to just governance? Ithink that would be really helpful for getting work pointed in the right direction getting the best people to be part of it. Ideally you'd come by and talk to them about progress once a month or whatever. We generically call people involved in some limited way ni YC "part-time partners" (we do that with Peter Thiei for exampie, though at this point he's very involved) but we could call ti whatever you want. Even fi you can't really spend time on ti but can be publicly supportive, that would still probably be really helpful for recruiting.
5. I think the right plan with the regulation letter is to wait for this to get going and then! can just release ti with a message like "now that we are doing this, I've been thinking a lot about what sort of constraints the world needs for safefy." Im' happy to leave you of as a signatory. Ialso suspect that after it's out more peopie will be willing to get behind it.
Sam
I missed something.
https://www.ycombinator.com/blog/welcome-peter
From the quoted text above. I.e. more than part time partner (heavily involved)
He's also part of the Paypal mafia along with Musk, btw.
https://www.bloomberg.com/opinion/articles/2024-03-01/openai...
I cleaned it up a bit but didn't notice that bug of 2 letters. I used Preview for macOS, for what it's worth. I also wonder why it swapped two letter words
The original had a `<!-|if IsupportLists]->[NUM]) <-[endif]>` for each bullet point which I found interesting, haven't seen that before in emails
Link to pdf: https://www.courthousenews.com/wp-content/uploads/2024/02/mu... (reference page 40, exhibit 2)
You can find the number of letters of the redacted text and then guess who they are. It's fun!
It's hard not to be a bit cynical about such an arrangement.
Back when the public at least somewhat bought the idea that SV was socially progressive and would use its massive accumulation of capital for the good of humanity.
Here's a hint: If you ever think "I can't trust anyone else with this", you are probably doing something wrong.
https://meta.discourse.org/t/help-us-to-test-the-html-pastin...