Evolving OpenAI's Structure
openai.com
openai.com
> Instead of our current complex capped-profit structure—which made sense when it looked like there might be one dominant AGI effort but doesn’t in a world of many great AGI companies—we are moving to a normal capital structure where everyone has stock. This is not a sale, but a change of structure to something simpler.
OpenAI is capturing most of the value in the space (generic LLM models), even though they have competitors who are beating them on price or capabilities.
I think OpenAI may be able to maintain this position at least for the medium term because of their name recognition/prominence and they are still a fast mover.
I also think the US is going to ban all non-US LLM providers from the US market soon for "security reasons."
What do you consider an "LLM provider"? Is it a website where you interact with a language model by uploading text or images? That definition might become too broad too quickly. Hard to ban.
everyone will roll over if all large public companies roll over (and they will)
https://www.theregister.com/2025/02/03/us_senator_download_c...
One of them will eventually pass given that OpenAI is also pushing for protection:
Their relationship with MS breaking down is a bad omen. I'm already seeing non-tech users who use "Copilot" because their spouse uses it at work. Barely knowing it's rebadged GPT. You think they'll switch when MS replaces the backend with e.g. Anthropic? No chance.
MS, Google and Apple and Meta have gigantic levers to pull and get the whole world to abandon OpenAI. They've barely been pulling them, but it's a matter of time. People didn't use Siri and Bixby because they were crap. Once everyone's Android has a Gemini button that's just as good as GPT (which it already is (it's better) for anything besides image generation), people are going to start pressing them. And good luck to OpenAI fighting that.
Well Trump is interested in tariffing movies and South Korea took DeepSeek off mobile app stores, so they certainly may try. But for high-end tasks, DeepSeek R1 671B is available for download, so any company with a VPN to download it and the necessary GPUs or cloud credits can run it. And for consumers, DeepSeek V3's distilled models are available for download, so anyone with a (~4 year old or newer) Mac or gaming PC can run them.
If the only thing keeping these companies valuations so high is banning the competition, that's not a good sign for their long-term value. If you have to ban the competition, you can't be feeling good about what you're making.
For what it's worth, I think GPT o3 and o1, Gemini 2.5 Pro and Claude 3.7 Sonnet are good enough to compete. DeepSeek R1 is often the best option (due to cost) for tasks that it can handle, but there are times where one of the other models can achieve a task that it can't.
But if the US is looking to ban Chinese models, then that could suggest that maybe these models aren't good enough to raise the funding required for newer, significantly better (and more expensive) models. That, or they just want to stop as much money as possible from going to China. Banning the competition actually makes the problem worse though, as now these domestic companies have fewer competitors. But I somewhat doubt there's any coherent strategy as to what they ban, tariff, etc.
Switching from ChatGPT to the many competitors is neither expensive nor painful.
OpenAI loses billions and is at the mercy of getting new investors to fund the losses. It has many plausible competitors.
and that makes complete sense if you don't have a lay person's understanding of the tech. Language models were never going to bring about "AGI."
This is another nail in the coffin
Which sounds pretty in-line with the SV culture of putting profit above all else.
If I were a person like several of the people working on AI right now (or really, just heading up tech companies), I could be the kind to look at a possible world-ending event happening in the next - eh, year, let's say - and just want to have a party at the end of the world.
Five years to ten years? Harder to predict.
The window there would at _least_ include the next 5 years, though obviously not ten.
It will likely require research breakthroughs, significant hardware advancement, and anything from a few years to a few decades. But it's coming.
ChatGPT was released 2.5 years ago, and look at all the crazy progress that has been made in that time. That doesn't mean that the progress has to continue, we'll probably see a stall.
But AIs that are on a level with humans for many common tasks is not that far off.
There's a lot of literature on this, and if you've been in the industry for any amount of time since the 1950s, you have seen at least one AI winter.
LLMs destroying any sort of capacity (and incentive) for the population to think pushes this further and further out each day
I don’t agree that this will affect ML progress much, since the general population isn’t contributing to core ML research.
Most HN people are probably too young to remember that the nanotech post-scarcity singularity was right around the corner - just some research and engineering way - which was the widespread opinion in 1986 (yes, 1986). It was _just as dramatic_ as today's AGI.
That took 4-5 years to fall apart, and maybe a bit longer for the broader "nanotech is going to change everything" to fade. Did nanotech disappear? No, but the notion of general purpose universal constructors absolutely is dead. Will we have them someday? Maybe, if humanity survives a hundred more years or more, but it's not happening any time soon.
There are a ton of similarities between nanotech-nanotech singularity and the moderns LLM-AGI situation. People point(ed) to "all the stuff happening" surely the singularity is on the horizon! Similarly, there was the apocalytpic scenario that got a ton of attention and people latching onto "nanotech safety" - instead of runaway AI or paperclip engines, it was Grey Goo (also coined in 1986).
The dynamics of the situation, the prognostications, and aggressive (delusional) timelines, etc. are all almost identical in a 1:1 way with the nanotech era.
I think we will have both AGI and general purpose universal constructors, but they are both no less than 50 years away, and probably more.
So many of the themes are identical that I'm wondering if it's a recurring kind of mass hysteria. Before nanotech, we were on the verge of genetic engineering (not _quite_ the same level of hype, but close, and pretty much the same failure to deliver on the hype as nanotech) and before that the crazy atomic age of nuclear everything.
Yes, yes, I know that this time is different and that AI is different and it won't be another round of "oops, this turned out to be very hard to make progress on and we're going to be in a very slow, multi-decade slow-improvement regime, but that has been the outcome of every example of this that I can think of.
It seems like nanotech is all around us now, but the term "nanotech" has been redefined to mean something different (larger scale, less amazing) from Drexler's molecular assemblers.
The problem is that the distance between a nano thin film or an interesting but ultimately rigid nano scale transistor and a programmable nano level sized robot is enormous, despite similar sizes. Same like the distance between an autocomplete heavily relying on the preexisting external validators (compilers, linters, static code analyzers etc.) and a real AI capable of thinking is equally enormous.
I thought this was a "we know we can't" thing rather than a "not with current technology" thing?
The idea of scaling up LLMs and hoping is .. pretty silly.
probably true but this statement would be true if when is 2308 which would defeat the purpose of the statement. when first cars started rolling around some mates around the campfire we saying “not if but when” we’ll have flying cars everywhere and 100 years later (with amazing progress in car manufacturing) we are nowhere near… I think saying “when, not if” is one of those statements that while probably indisputable in theory is easily disputable in practice. give me “when” here and I’ll put up $1,000 to a charity of your choice if you are right and agree to do the same thing if wrong
To begin with, systems that don't tell people to use elmer's glue to keep the cheese from sliding off the pizza, displaying a fundamental lack of understanding of.. everything. At minimum it needs to be able to reliably solve hard, unique, but well-defined problems like a group of the most cohesive intelligent people could. It's certainly not AGI until it can do a better job than the most experienced, talented, and intelligent knowledge workers out there.
Every major advancement (which LLMs certainly are) has caused some disruption in the fields it affected, but that isn't useful criteria that can differentiate between "crude but useful tool" from "AGI".
I mean "hard" in the sense that it can reliably replace the best software developers, civil engineers, lawyers, diagnosticians. Not just in economic sense, but by reliably matching the quality of their work 100% of the time.
It should be capable of methodically and reliably arriving at correct answers without expert intervention. It shouldn't be the case that some people claim that they don't know how to code and the LLM generated an entire project for them, while I can confidently claim that LLMs fall flat on their face almost every time I try to use them for more delicate business logic.
you can see a pattern of fairly steady progress in different aspects, like they matched humans for image recognition around 2015 but 'complex reasoning' is still much worse than humans but rising.
Looking at the graph, I'd guess maybe five years before it can do all human skills which is roughly AGI?
I've got a personal AGI test of being able to fix my plumbing, given a robot body. Which they are way off just now.
It has taken tens to hundred of billions of dollars without equivalent economic justification(yet) before to reach here. I am not saying economic justification doesn't exist or wont come in the future, just that the upfront investment and risk is already in order of magnitude of what the largest tech companies can expend.
If the the next generation requires hundreds of billions or trillions [2] upfront and a very long time to make returns, no one company (or even country) could allocate that kind of resources.
Many cases of such economically limited innovations[1], nuclear fusion is the classic always 20 years away example. Another close one is anything space related, we cannot replicate in next 5 years what we already achieved from 50 years ago of say landing on the moon and so on.
From a just a economic perspective it is a definitely a "If", without even going into the technology challenges.
[1]Innovations in cost of key components can reshape economics equation, it does happen (as with spaceX) but it also not guaranteed like in fusion.
[2] The next gen may not be close enough to AGI. AGI could require 2-3 more generations ( and equivalent orders of magnitude of resources), which is something the world is unlikely to expend resources on even if it had them.
We have zero evidence for this. (Folks said the same shit in the 80s.)
I want to believe, man.
Yeah; and:
We want to open source very capable models.
https://pdfernhout.net/on-funding-digital-public-works.html#...
"Consider this way of looking at the situation. A 501(c)3 non-profit creates a digital work which is potentially of great value to the public and of great value to others who would build on that product. They could put it on the internet at basically zero cost and let everyone have it effectively for free. Or instead, they could restrict access to that work to create an artificial scarcity by requiring people to pay for licenses before accessing the content or making derived works. If they do the latter and require money for access, the non-profit can perhaps create revenue to pay the employees of the non-profit. But since the staff probably participate in the decision making about such licensing (granted, under a board who may be all volunteer), isn't that latter choice still in a way really a form of "self-dealing" -- taking public property (the content) and using it for private gain? From that point of view, perhaps restricting access is not even legal?"
"Self-dealing might be clearer if the non-profit just got a grant, made the product, and then directly sold the work for a million dollars to Microsoft and put the money directly in the staff's pockets (who are also sometimes board members). Certainly if it was a piece of land being sold such a transaction might put people in jail. But because the content or software sales are small and generally to their mission's audience they are somehow deemed OK. The trademark-infringing non-profit-sheltered project I mention above is as I see it in large part just a way to convert some government supported PhD thesis work and ongoing R&D grants into ready cash for the developers. Such "spin-offs" are actually encouraged by most funders. And frankly if that group eventually sells their software to a movie company, say, for a million dollars, who will really bat an eyebrow or complain? (They already probably get most of their revenue from similar sales anyway -- but just one copy at a time.) But how is this really different from the self-dealing of just selling charitably-funded software directly to Microsoft and distributing a lump sum? Just because "art" is somehow involved, does this make everything all right? To be clear, I am not concerned that the developers get paid well for their work and based on technical accomplishments they probably deserve that (even if we do compete for funds in a way). What I am concerned about is the way that the proprietary process happens such that the public (including me) never gets full access to the results of the publicly-funded work (other than a few publications without substantial source)."
That said, charging to provide a service that costs money to supply (e.g. GPU compute) is not necessarily self-dealing. It is restricting the source code or using patents to create artificial scarcity around those services that could be seen that way.
Your 2001 essay isn't a good parallel to OpenAI's situation.
OpenAI wasn't "publicly funded" i.e. with public donations or government grants.
The non-profit was started and privately funded by a small group of billionaires and other wealthy people (Elon Musk donates $44 million, Reid Hoffman, etc collectively pledging $1 billion of their own money).
They miscalculated in thinking their charity donations would be enough to recruit the PhD machine learning researchers and pay the high GPU costs to create the AI alternative to Google DeepMind, etc. Their 2015 assumptions about future AI development costs were massively underestimated and now they look like bad for trying to convert it to a for-profit enterprise. Instead of a big conversion to for-profit, they now will settle with keeping a subsidiary that's for-profit. Somewhat like other entities structured as a non-profit that owns for-profit subsidiaries such as Mozilla, Girl Scouts, Novo Nordisk, etc.
Obviously with hindsight... if they had to do it all over, they would just create the reverse structure of creating the OpenAI for-profit company as the "parent entity" that pledges to donate money to charities. E.g. Amazon Inc is the for-profit that donates to Housing Equity Fund for affordable housing.
Taxes are on profits not revenue. The for-profit OpenAI LLC subsidiary created in 2019 would have been the entity that owes taxes but it has been losing money and never made any profits to tax.
Yesterday's news about switching from for-profit LLC to for-profit PBC still leaves a business entity that's liable for future taxes on profits.
Not only is there infinite incentive to compete, but theres decreasing costs to. The only world in which AGI is winner take all is a world in which it is extremely controlled to the point at which the public cant query it.
I wonder, do you have a hypothesis as to what would be a measurement that would differentiate AGI vs Not-AGI?
So one fundamental difference is that AGI would not need some absurdly massive data dump to become intelligent. In fact you would prefer to feed it as minimal a series of the most primitive first principles as possible because it's certain that much of what we think is true is going to end up being not quite so -- the same as for humanity at any other given moment in time.
We could derive more basic principles, but this one is fundamental and already completely incompatible with our current direction. Right now we're trying to essentially train on the entire corpus of human writing. That is a defacto acknowledgement that the absolute endgame for current tech is simple mimicry, mistakes and all. It'd create a facsimile of impressive intelligence because no human would have a remotely comparable knowledge base, but it'd basically just be a glorified natural language search engine - frozen in time.
If you took the average human from birth and gave them only 'the most primitive first principles', the chance that they would have novel insights into medicine is doubtful.
I also disagree with your following statement:
> Right now we're trying to essentially train on the entire corpus of human writing. That is a defacto acknowledgement that the absolute endgame for current tech is simple mimicry
At worst it's complex mimicry! But I would also say that mimicry is part of intelligence in general and part of how humans discover. It's also easy to see that AI can learn things - you can teach an AI a novel language by feeding in a fairly small amount of words and grammar of example text into context.
I also disagree with this statement:
> One fundamental difference is that AGI would not need some absurdly massive data dump to become intelligent
I don't think how something became intelligent should affect whether it is intelligent or not. These are two different questions.
You didn't teach it, the model is still the same after you ran that. That is the same as a human following instructions without internalizing the knowledge, he forgets it afterward and didn't learn what he performed. If that was all humans did then there would be no point in school etc, but humans do so much more than that.
As long as LLM are like an Alzheimer's human they will never become a general intelligence. And following instructions is not learning at all, learning is building an internal model for those instructions that is more efficient and general than the instructions themselves, humans do that and that is how we manage to advance science and knowledge.
Then when OpenAI does another training run it can also internalise that knowledge into the weights.
This is much like humans - we have short term memory (where it doesn't get into the internal model) and then things get baked into long term memory during sleep. AI's have context-level memory, and then that learning gets baked into the model during additional training.
Although whether or not it changed the weights IMO is not a prerequisite for whether something can learn something or not. I think we should be able to evaluate if something can learn by looking at it as a black-box, and we could make a black-box which would meet this definition if you spoke to a LLM and limited it to it's max context length each day, and then ran an overnight training run to incorporate learned knowledge into weights.
> So one fundamental difference is that AGI would not need some absurdly massive data dump to become intelligent.
The first 22 years of life for a “western professional adult” is literally dedicated to a giant bootstrapping info dump
The zero training version not only ended up dramatically outperforming the 'expert' version, but reached higher levels of competence exponentially faster. And that should be entirely expected. There were obviously tremendous flaws in our understanding of the game, and training on those flaws resulted in software seemingly permanently handicapping itself.
Minimal expert training also has other benefits. The obvious one is that you don't require anywhere near the material and it also enables one to ensure you're on the right track. Seeing software 'invent' fundamental arithmetic is somewhat easier to verify and follow than it producing a hundred page proof advancing, in a novel way, some esoteric edge theory of mathematics. Presumably it would also require orders of magnitude less operational time to achieve such breakthroughs, especially given the reduction in preexisting state.
The moment after human birth the human agent starts a massive information gathering process - that no other system really expects much output from in a coherent way - for 5-10 years. Aka “data dump” some of that data is good, and some of it is bad. This in turn leads to biases, it leads to poor thinking models; everything that you described, is also applicable to every intelligent system - including humans. So again you presupposing that there’s some kind of perfect information benchmark that couldn’t exist.
When that system comes out of the birth canal it already has embedded in it millions of years of encoded expectations predictability systems and functional capabilities that are going to grow independent of what the environment does (but will be certainly shaped in its interactions by the environment).
So no matter what, you have a structured system of interaction that must be loaded with previously encoded data (experience, transfer learning etc) with and it doesn’t matter what type of intelligent system you’re talking about there are foundational assumptions at the physical interaction layer that encode all previous times steps of evolution.
Said an easier way: a lobster, because of the encoded DNA that created it, will never have the same capabilities as a human, because it is structured to process information completely differently and their actuators don’t have the same type and level of granularity as human actuators.
Now assume that you are a lobster compared to a theoretical AGI in sensor-effector combination. Most likely it would be structured entirely differently than you are as a biological thing - but the mere design itself carries with it an encoding of structural information of all previous systems that made it possible.
So by your definition you’re describing something that has never been seen in any system and includes a lot of assumptions about how alternative intelligent systems could work - which is fair because I asked your opinion.
The next time your in the wilds, it's quite amazing to consider that your ancestors - millennia past, would have looked at, more or less, these exact same wilds but with so much less knowledge. Yet nonetheless they would discover such knowledge - teaching themselves, and ourselves, to build rockets, put a man on the Moon, unlock the secrets of the atom, and so much more. All from zero.
---
What your example and elaboration focus on is the nature of intelligence, and the difficulty in replicating it. And I agree. This is precisely we want to avoid making the problem infinitely more difficult, costly, and time consuming by dumping endless amounts of knowledge in the equation.
I’m curious what epistemological grounding you are basing your claim on
EDIT: There can be levels of AGI. Google DeepMind have proposed a framework that would classify ChatGPT as "Emerging AGI".
I don't know if it is optimism or delusions of grandeur that drives people to make claims like AGI will be here in the next decade. No, we are not getting that.
And what do you think would happen to us humans if such AGI is achieved? People's ability to put food on the table is dependent on their labor exchanged for money. I can guarantee for a fact, that work will still be there but will it be equitable? Available to everyone? Absolutely not. Even UBI isn't going to cut it because even with UBI people still want to work as experiments have shown. But with that, there won't be a majority of work especially paper pushing mid level bs like managers on top of managers etc.
If we actually get AGI, you know what would be the smartest thing for such an advanced thing to do? It would probably kill itself because it would come to the conclusion that living is a sin and a futile effort. If you are that smart, nothing motivates you anymore. You will be just a depressed mass for all your life.
That's just how I feel.
There can be levels of AGI. Google DeepMind have proposed a framework that would classify ChatGPT as "Emerging AGI".
ChatGPT can solve problems that it was not explicitly trained to solve, across a vast number of problem domains.
https://arxiv.org/pdf/2311.02462
The paper is summarized here https://venturebeat.com/ai/here-is-how-far-we-are-to-achievi...
Think about it - the original definition of AGI was basically a machine that can do absolutely anything at a human level of intelligence or better.
That kind of technology wouldn't just appear instantly in a step change. There would be incremental progress. How do you describe the intermediate stages?
What about a machine that can do anything better than the 50th percentile of humans? That would be classified as "Competent AGI", but not "Expert AGI" or ASI.
> fancy search engine/auto completer
That's an extreme oversimplification. By the same reasoning, so is a person. They are just auto completing words when they speak. No that's not how deep learning systems work. It's not auto complete..
It's really not. The Space Shuttle isn't an emerging interstellar spacecraft, it's just a spacecraft. Throwing emerging in front of a qualifier to dilute it is just bullshit.
> By the same reasoning, so is a person. They are just auto completing words when they speak.
We have no evidence of this. There is a common trope across cultures and history of characterising human intelligence in terms of the era's cutting-edge technology. We did it with steam engines [1]. We did it with computers [2]. We're now doing it with large language models.
[1] http://metaphors.iath.virginia.edu/metaphors/24583
[2] https://www.frontiersin.org/journals/ecology-and-evolution/a...
The General Intelligence part of AGI refers to its ability to solve problems that it was not explicitly trained to solve, across many problem domains. We already have examples of the current systems doing exactly that - zero shot and few shot capabilities.
> We have no evidence of this.
That's my point. Humans are not "autocompleting words" when they speak.
No, it's bringing something out of scope into the definition. Gluten-free means free of gluten. Gluten-free bagel verus sliced bread is a refinement--both started out under the definition. Glutinous bread, on the other hand, is not gluten free. As a result, "almost gluten free" is bullshit.
> That's my point. Humans are not "autocompleting words" when they speak
Humans are not. LLMs are. It turns out that's incredibly powerful! But it's also limiting in a way that's fundamentally important to the definition of AGI.
LLMs bring us closer to AGI in the way the inventions of writing, computers and the internet probably have. Calling LLMs "emerging AGI" pretends we are on a path to AGI in a way we have zero evidence for.
Bad analogy. That's a binary classification. AGI systems can have degrees of performance and capability.
> Humans are not. LLMs are.
My point is that if you oversimplify LLMs to "word autocompletion" then you can make the same argument for humans. It's such an oversimplification of the transformer / deep learning architecture that it becomes meaningless.
The "g" in AGI requires the AI be able to perform "the full spectrum of cognitively demanding tasks with proficiency comparable to, or surpassing, that of humans" [1]. Full and not full are binary.
> if you oversimplify LLMs to "word autocompletion" then you can make the same argument for humans
No, you can't, unless you're pre-supposing that LLMs work like human minds. Calling LLMs "emerging AGI" pre-supposes that LLMs are the path to AGI. We simply have no evidence for that, no matter how much OpenAI and Google would like to pretend it's true.
[1] https://en.wikipedia.org/wiki/Artificial_general_intelligenc...
> No, you can't, unless you're pre-supposing that LLMs work like human minds.
You are missing the point. If you reduce LLMs to "word autocompletion" then you completely ignore the the attention mechanism and conceptual internal representations. These systems have deep learning models with hundreds of layers and trillions of weights. If you completely ignore all of that, then by the same reasoning (completely ignoring the complexity of the human brain) we can just say that people are auto-completing words when they speak.
Sure, Google wants to redefine AGI so it looks like things that aren’t AGI can be branded as such. That definition is, correctly in my opinion, being called out as bullshit.
> obviously there will be stages in between
We don’t know what the stages are. Folks in the 80s were similarly selling their expert systems as a stage to AGI. “Emerging AGI” is a bullshit term.
> If you reduce LLMs to "word autocompletion" then you completely ignore the the attention mechanism and conceptual internal representations. These systems have deep learning models with hundreds of layers and trillions of weights
Fair enough, granted.
It is not a redefinition. It's a classification for AGI systems. It's a refinement.
Other researchers are also trying to classify AGI systems. It's not just Google. Also, there is no universally agreed definition of AGI.
> We don’t know what the stages are. Folks in the 80s were similarly selling their expert systems as a stage to AGI. “Emerging AGI” is a bullshit term.
Generalization is a formal concept in machine learning. There can be degrees of generalized learning performance. This is actually measurable. We can compare the performance of different systems.
The g in AGI is General. I don't what world you think Generality isn't a spectrum, but it's sure as hell isn't this one.
"A framework for classifying AGI by performance and autonomy was proposed in 2023 by Google DeepMind researchers. They define five performance levels of AGI: emerging, competent, expert, virtuoso, and superhuman"
In the second paragraph:
"Some researchers argue that state‑of‑the‑art large language models already exhibit early signs of AGI‑level capability, while others maintain that genuine AGI has not yet been achieved."
The entire article makes it clear that the definitions and classifications are still being debated and refined by researchers.
Edit: because if "AGI" doesn't mean that... then what means that and only that!?
"Agentic AI" means that.
Well, to some people, anyway. And even then, people are already arguing about what counts as agency.
That's the trouble with new tech, we have to invent words for new stuff that was previously fiction.
I wonder, did people argue if "horseless carriages" were really carriages? And "aeroplane" how many argued that "plane" didn't suit either the Latin or Greek etymology for various reasons?
We never did rename "atoms" after we split them…
And then there's plain drift: Traditional UK Christmas food is the "mince pie", named for the filling, mincemeat. They're usually vegetarian and sometimes even vegan.
It's kind of a simple enough concept... it's really just something that functions on par with how we do. If you've built that, you've built AGI. If you haven't built that, you've built a very capable system, but not AGI.
"Can", but not "must". The difference between an LLM being harnessed to be a customer service agent, or a code review agent, or a garden planning agent, can be as little as the prompt.
And in any case, the point was that the concept of "completely autonomous agentic intelligence capable of operating on long-term planning horizons" is better described by "agentic AI" than by "AGI".
> It's kind of a simple enough concept... it's really just something that functions on par with how we do.
"On par with us" is binary thinking — humans aren't at the same level as each other.
The problem we have with LLMs is the "I"*, not the "G". The problem we have with AlphaGo and AlphaFold is the "G", not the ultimate performance (which is super-human, an interesting situation given AlphaFold is a mix of Transformer and Diffusion models).
For many domains, getting a degree (or passing some equivalent professional exam) is just the first step, and we have a long way to go from there to being trusted to act competently, let alone independently. Someone who started a 3-year degree just before ChatGPT was released, will now be doing their final exams, and quite a lot of LLMs operate like they have just about scraped through degrees in almost everything — making them wildly superhuman with the G.
The G-ness of an LLM only looks bad when compared to all of humanity collectively; they are wildly more general in their capabilities than any single one of us — there are very few humans who can even name as many languages as ChatGPT speaks, let alone speak them.
* they need too many examples, only some of that can be made up for by the speed difference that lets machines read approximately everything
Stepping back for a moment - do we actually want something that has agency?
Name me a human that also doesn't need direction or guidance to do a task, at least one they haven't done before
Literally everything that's been invented.
To be fair, there is a section of the population whose useful intelligence can roughly be summed up as that or worse.
- Processing visual data and classifying objects within their field of vision.
- Processing auditory data, identifying audio sources and filtering out noise.
- Maintaining an on-going and continuous stream of thoughts and emotions.
- Forming and maintaining complex memories on long-term and short-term scales.
- Engaging in self-directed experimentation or play, or forming independent wants/hopes/desires.
I could sit here all day and list the forms of intelligence that humans and other intelligent animals display which have no obvious analogue in an AI product. It's true that individual AI products can do some of these things, sometimes better than humans could ever, but there is no integrated AGI product that has all these capabilities. Let's give ourselves a bit of credit and not ignore or flippantly dismiss our many intelligent capabilities as "useless."
No, I’m using useful problem solving as my benchmark. There are useless forms of intelligence. And that’s fine. But some people have no useful intelligence and show no evidence of the useless kind. They don’t hit any of the bullets you list, there just isn’t that curiosity and drive and—I suspect—capacity to comprehend.
I don’t think it’s intrinsic. I’ve seen pets show more curiosity than some folk. But due to nature and nurture, they just aren’t intelligent to any material stretch.
The turing test was succesfull. Pre chatGPT, I would not have believed, that will happen so soon.
LLMs ain't AGI, sure. But they might be an essential part and the missing parts maybe already found, just not put together.
And work there will be always plenty. Distributing ressources might require new ways, though.
The very people whose theories about language are now being experimentally verified by LLMs, like Chomsky, have also been discrediting the Turing test as pseudoscientific nonsense since early 1990s.
It's one of those things like the Kardashev scale, or Level 5 autonomous driving, that's extremely easy to define and sounds very cool and scientific, but actually turns out to have no practical impact on anything whatsoever.
Bots, that are now allmost indistinguishable from humans, won't have a practical impact? I am sceptical. And not just because of scammers.
In particular we redefined the test to make it passable. In Turing's original concept the competent investigator and participants were all actively expected to collude against the machine. The entire point is that even with collusion, the machine would be able to pass. Instead modern takes have paired incompetent investigators alongside participants colluding with the machine, probably in an effort to be part 'of something historic'.
In "both" (probably more, referencing the two most high profile - Eugene and the large LLMs) successes, the interrogators consistently asked pointless questions that had no meaningful chance of providing compelling information - 'How's your day? Do you like psychology? etc' and the participants not only made no effort to make their humanity clear, but often were actively adversarial obviously intentionally answering illogically, inappropriately, or 'computery' to such simple questions. And the tests are typically time constrained by woefully poor typing skills (this the new normal in the smartphone gen?) to the point that you tend to get anywhere from 1-5 interactions of a few words each.
The problem with any metric for something is that it often ends up being gamed to be beaten, and this is a perfect example of that.
And I did not looked into it (I also don'think the test has too much relevance), but fooling the average person sounds plausible by now.
Now sounding plausible is what LLMs are optimized for and not being plausible, still, I would not have thought we get so far so quick 10 years ago. So I am very hesistant about the future.
The two concepts have historically been inexorably linked in sci-fi, which will likely make the first AGI harder to recognize as AGI if it lacks consciousness, but I'd argue that simple "unconscious AGI" would be the superior technology for current and foreseeable needs. Unconscious AGI can be employed purely as a tool for massive collective human wealth generation; conscious AGI couldn't be used that way without opening a massive ethical can of worms, and on top of that its existence would represent an inherent existential threat.
Conscious AGI could one day be worthwhile as something we give birth to for its own sake, as a spiritual child of humanity that we send off to colonize distant or environmentally hostile planets in our stead, but isn't something I think we'd be prepared to deal with properly in a pre-post-scarcity society.
It isn't inconceivable that current generative AI capabilities might eventually evolve to such a level that they meet a practical bar to be considered unconscious AGI, even if they aren't there yet. For all the flak this tech catches, it's easy to forget that capabilities which we currently consider mundane were science fiction only 2.5 years ago (as far as most of the population was concerned). Maybe SOTA LLMs fit some reasonable definition of "emerging AGI", or maybe they don't, but we've already shifted the goalposts in one direction given how quickly the Turing test became obsolete.
Personally, I think current genAI is probably a fair distance further from meeting a useful definition of AGI than those with a vested interest in it would admit, but also much closer than those with pessimistic views of the consequences of true AGI tech want to believe.
It isn't close at all.
It is not hard to imagine a "cooking robot" as a black box that — given the appropriate ingredients — would cook any dish for you. Press a button, say what you want, and out it comes.
Internally, the machine would need to perform lots of tasks that we usually associate with intelligence, from managing ingredients and planning cooking steps, to fine-grained perception and manipulation of the food as it is cooking. But it would not be conscious in any real way. Order comes in, dish comes out.
Would we use "intelligent" to describe such a machine? Or "magic"?
A machine could be super intelligent at solving real world practical tasks, better than any human, without being conscious.
We don't have a proper definition of consciousness. Consciousness is infinitely more mysterious than measurable intelligence.
I don't think there has ever been a time in history when work has been equitable and available to everyone.
Of course, that isn't to say that AI can't make it worse then it is now.
"AGI" was already a goalpost move from "AI" which has been gobbled up by the marketing machine.
This is current research. The classification of AGI systems is currently being debated by AI researchers.
It's a classification system for AGI, not a redefinition. It's a refinement.
Also there is no universally accepted definition of AGI in the first place.
Here is a mainstream opinion about why AGI is already here. Written by one of the authors the most widely read AI textbook: Artificial Intelligence: A Modern Approach https://en.wikipedia.org/wiki/Artificial_Intelligence:_A_Mod...
Can ChatGPT drive a car? No, we have specialized models for driving vs generating text vs image vs video etc etc. Maybe ChatGPT could pass a high school chemistry test but it certainly couldn't complete the lab exercises. What we've built is a really cool "Algorithm for indexing generalized data", so you can train that Driving model very similarly to how you train the Text model without needing to understand the underlying data that well.
The author asserts that because ChatGPT can generate text about so many topics that it's general, but it's really only doing 1 thing and that's not very general.
I think we need to separate the thinking part of intelligence from tool usage. Not everyone can use every tool at a high level of expertise.
but so do I!
Of course they can. We already have computer controlled car systems, the reason LLMs aren't used to drive them is because AI systems that specialize in text are a poor choice for driving - specialized driving models will always outperform them for a variety of technical reasons.
That was my whole point. Maybe in theory an LLM could learn to drive a car, but they can't today because they don't physically have access to cars they could try to drive just like a person who can't learn to use a tool because they're physically limited from using it.
Likewise for "intelligent", and even "artificial".
So no, ChatGPT can't drive a car*. But it knows more about car repairs, defensive driving, global road features (geoguesser), road signs in every language, and how to design safe roads, than I'm ever likely to.
* It can also run python scripts with machine vision stuff, but sadly that's still not sufficient to drive a car… well, to drive one safety, anyway.
How about we have ChatGPT start with a simple task like reliably generating JSON schema when asked to.
Hint: it will fail.
This doesn’t imply that it’s ideal for driving cars, but to say that it’s not capable of driving general intelligence is incorrect in my view.
Same model trained on audio, video, images, text - not separate specialized components stitched together.
Last time I checked, in an Anthropic paper, they asked the model to count something. They examined the logits and a graph showing how it arrived at the answer. Then they asked the model to explain its reasoning, and it gave a completely different explanation, because that was the most statistically probable response to the question. Does that seem like AGI to you?
There’s a good reason why schools spend so much time training that skill!
It is easy to see why, since the LLM doesn't communicate what it thinks it communicates what it thinks a human would communicate. A human would explain their inner process, and then go through that inner process. An LLM would explain a humans inner process, and then generate a response using a totally different process.
So while its true that humans doesn't have perfect introspection, the fact that we have introspection about our own thoughts at all is extremely impressive. An LLM has no part that analyzes its own thoughts the way humans do, meaning it has no clue how it thinks.
I have no idea how you would even build introspection into an AI, like how are we able to analyze our own thoughts? What is even a thought? What would this introspection part of an LLM do, what would it look like, would it identify thoughts and talk about them the way we do? That would be so cool, but that is not even on the horizon, I doubt we will ever see that in our lifetime, it would need some massive insight changing the AI landscape at its core to get there.
But, once you have that introspection I think AGI will happen almost instantly. Currently we use dumb math to train the model, that introspection will let the model train itself in an intelligent way, just like humans do. I also think it will never fully replace humans without introspection, intelligent introspection seems like a fundamental part to general intelligence and learning from chaos.
The first-mover advantages of an AGI that can improve itself are theoretically unsurmountable.
But OpenAI doesn't have a path to AGI any more than anyone else. (It's increasingly clear LLMs alone don't make the cut.) And the market for LLMs, non-general AI, is very much not winner takes all. In this announcement, OpenAI is basically acknowledging that it's not getting to self-improving AGI.
This has some baked assumptions about cycle time and improvement per cycle and whether there's a ceiling.
To be precise, it assumes a low variability in cycle time and improvement per cycle. If everyone is subjected to the same limits, the first-mover advantage remains insurmountable. I’d also argue that whether there is a ceiling matters less than how high it is. If the first AGI won’t hit a ceiling for decades, it will have decades of fratricidal supremacy.
How steeply the diminishing returns curve off at.
The most advanced tools are (and will continue to be) at a higher level of the stack, combining the leading models for different purposes to achieve results that no single provider can match using only their own models.
I see no reason to think this won't hold post-AGI (if that happens). AGI doesn't mean capabilities are uniform.
It does have some weasel words around value-aligned and safety-conscious which they can always argue but this could get interesting because they've basically agreed not to compete. A fairly insane thing to do in retrospect.
"Instead of our current complex non-competing structure—which made sense when it looked like there might be one dominant AGI effort but doesn’t in a world of many great AGI companies—we are moving to a normal competing structure where ..." is all it takes
That's always been pretty overtly the winner-take-all AGI scenario.
That's literally the premise of venture capital. This is a scenario where we're assuming ALL our bets will go to zero, except one which will be worth trillions. In that case you should bet on everything.
It's only in the opposite scenario (where every bet pays off with varying ROI) that it makes sense to go all-in on whichever bet seems most promising.
One remarkable advantage of being a "Public Benefit Corporation" is this it:
> prevent[s] shareholders from using a drop in stock value as evidence for dismissal or a lawsuit against the corporation[1]
In my view, it is their own shareholders that the directors of OpenAI are insulating themselves against.
Please promise to come back to this comment in 2030 and playfully mock me for ever being worried and I will buy you a coffee. If AGI is invented before 2030 please buy me one and let me mock you playfully.
"Open AI for-profit LLC will become a Public Benefit Corporation (PBC)"
followed by: "Profit cap is hereby removed" and finally "The Open AI non-profit will continue to control the PBC. We intend it to be a significant shareholder of the PBC."
This is originally from The Art of War.
We need to get closer to the norm and give shares of a for-profit to employees in order to create retention.
― Sun Tzu
Quite the arc from the original organization.
The primary difference is the observability - with satellites we had some confidence that other nations respected treaties, or that they had enough reaction time for mutual destruction, but with this AI development we lack all that.
We can selectively ban uses without banning the technology wholesale; e.g., nuclear power generation is permitted, while nuclear weapons are strictly controlled.
If it's winner takes all for the first company/nation to have AGI (presuming we can control it), then slowing down progress of any kind with regulation is a risk.
I don't think there's a good enough analogy to be made, like your nuclear power/weapons example.
The hypothetical benefits of an aligned AGI outweigh those of any other technology by orders of magnitude.
We should not be racing ahead because China is, but investing energy in alignment research and international agreements.
We do know that. By literally looking at China.
> The hypothetical benefits of an aligned AGI outweigh those of any other technology by orders of magnitude.
AGI aligned with whom?
I think the more relevant question is: Do you want to live in a Chinese dystopia, or a European one?
A non-AI dystopia is the least likely scenario.
With US already having lost ideologigal war with russia and China, Europe is very much next
No - I'm suggesting that China will reap the benefits of AI much more than Europe will, and they will eclipse Europe economically. Their dominance will follow, and they'll be able to dictate terms to other countries (just as the US is doing, and has been doing).
> And I don't think China will become a utopia with unregulated AI.
Did you miss all the places I used the word "dystopia"?
> My impression after having visited it was not one of a utopia, and knowing how they use technology, I don't think AI will usher it in, because our visions of utopia are at odds. They may well enjoy what they have.
Comparing China when I was a kid, not that long ago, to what it is now: It is a dystopia, and that dystopia is responsible for much of the improvements they've made. Enjoying what they have doesn't mean it's not a dystopia. Most people don't understand how willing humans are to live in a dystopia if it improves their condition significantly (not worrying too much about food, shelter, etc).
No, just control. America exerts influence and control over Europe without having had to attack it in generations.
Right now it's operated by a bunch of people who think that you can directly relate the amount of money a venture could make in the next 90 days to its net benefit for society. Government telling them how they can and cannot make that money, in their minds, is government telling them that they cannot bring maximum benefit to society.
Now, is this mindset myopic to everything that most people have in their lived experience? Is it ethically bankrupt and held by people who'd sell their own mothers for a penny if they otherwise couldn't get that penny? Would those people be banished to a place beyond human contact for the rest of their existence by functioning organs of an even somewhat-sane society?
I don't know. I'm just asking questions.
It might not be a direct US-govt project like the Manhattan Project was, but it doesn't have to. The government has the ties it needs with the heads of all these AI companies, and if it comes to it, the US-govt has the muscle and legal authority to reign control over it.
A good deal for everyone involved really. These companies get to make bank and technology that furthers their market dominance, the US-govt gets potentially "Manhattan project"-level pivotal technology— it's elites helping elites.
Mostly OpenAI and DeepMind and it stunk of 'pulling up the drawbridge behind them' and pivoting from actual harm to theoretical harm.
For a crowd supposedly entrenched in startups, it's amazing everyone here is so slow to recognise it's all funding pitches and contract bidding.
The "digital god" angle might explain why. For many, this has become a religious movement, a savior for an otherwise doomed economic system.
Omnipotent deities can never be held responsible for famine and natural disasters ("God has a plan for us all"). AI currently has the same get-out-of-jail free card where mistakes that no literate human would ever make are handwaved away as "hallucinations" that can be exorcised with a more sophisticated training model ("prayers").
Id love to believe there is more to life than the AI future, or that we as humans are destined to be perpetually happy and live meaningful. However I currently dont see how our current levels of extreme prosperity are anything more than an evolutionary blip, even if we could make them last several millennia more.
Sounds like payola for the enterprising and experienced mercenary.
But OpenAI isn't limited to creating LLMs. OpenAI's objective is not to create LLMs but to create artificial general intelligence that is better than humans at all intellectual tasks. Examples of such tasks include:
1. Designing nuclear weapons.
2. Designing and troubleshooting mining, materials processing, and energy production equipment.
3. Making money by investing in the stock market.
4. Discovering new physics and chemistry.
5. Designing and troubleshooting electronics such as GPUs.
6. Building better AI.
7. Cracking encryption.
8. Finding security flaws in computer software.
9. Understanding the published scientific literature.
10. Inferring unpublished discoveries of military significance from the published scientific literature.
11. Formulating military strategy.
Presumably you can see that a system capable of doing all these things can easily be used to produce an unlimited quantity of nuclear weapons, thus making it more powerful than any nuclear weapon.
If LLMs turn out not to be able to do those things better than humans, OpenAI will try other approaches, sooner or later. Maybe it'll turn out to be impossible, or much further off than expected, but that's not what OpenAI is claiming.
You are surely correct that there are weaker imaginable AIs than the strongly superhuman AI that OpenAI and I are talking about which would still be more powerful than nuclear weapons, but they are more debatable. For example, whether discovering new physics would permit the construction of new, more powerful weapons is debatable; it didn't help Archimedes or Tipu Sultan. So discussing such weak claims is likely to end up off in the weeds of logistics and speculation about exactly what kind of undiscovered physics and math would come to light. Instead, I focused on the most obviously correct ways that strongly superhuman AI would be more powerful than nuclear weapons.
These may not be the most practically important ways. Maybe any strongly superhuman AI would immediately discover a way to explode the sun, or to control people's minds, or to build diamondoid molecular nanotechnology, or to genetically engineer super-plagues, or to collapse the false vacuum. Any of those would make nuclear weapons seem insignificant. But claims like those are much more uncertain than the very simple question before us: whether what OpenAI is trying to develop would be more powerful than nuclear weapons. Obviously it would be, by my reasoning in the grandparent comment, even if this isn't a false vacuum, if the sticky fingers problem makes diamondoid nanotechnology impossible, if people's minds are inherently uncontrollable, etc. So we don't need to resolve those other, more difficult questions in order to do the much easier task of ranking OpenAI's objective relative to nuclear weapons.
LLMs are great at making you think they are the other but aren't.
The questions you are bringing up about the possible limits of the LLM approach are interesting open research questions, and while I really doubt your implicit claim to have resolved them, they are ultimately irrelevant to the topic at hand, which, I will remind you, is the astounding novelty of the situation where
> many companies operating in the public eye are basically stating "We are creating a digital god, an instrument more powerful than any nuclear weapon" and raising billions to do it, and nobody bats an eye...
Note that there is nothing about LLMs in this proposition, and the particular company we're implicitly most focused on—OpenAI—has already developed a number of well-known models that aren't LLMs and plans to keep doing so.
The intersection of the two seems to be quite hard to find.
At the state that we're in the AIs we're building are just really useful input/output devices that respond to a stimuli (e.g., a "prompt"). No stimuli, no output.
This isn't a nuclear weapon. We're not going to accidentally create Skynet. The only thing it's going to go nuclear on is the market for jobs that are going to get automated in an economy that may not be ready for it.
If anything, the "danger" here is that AGI is going to be a printing press. A cotton gin. A horseless carriage -- all at the same time and then some, into a world that may not be ready for it economically.
Progress of technology should not be artitrarily held back to protect automateable jobs though. We need to adapt.
So you don't mind if your economic value drops to zero, with all human labour replaced by machines?
Dependent on UBI, existing in a basic pod, eating rations of slop.
This was the fear when the cotton gin was invented. It was the ear when cars were created. The same complaint happened with the introduction of electronic, automated, telephone switchboards.
Jobs change. Societies change. Unemployment worldwide, is near the lowest it has ever been. Work will change. Society will eventually move to a currency based on energy production, or something equally futuristic.
This doesn't mean that getting there will be without pain.
The economic value of human labour will drop to zero. That would be an existential threat to our civilization.
There's so much to do, explore and learn. The prospect of AI stealing my job is only scary because my income depends on this job.
Hobbies, hanging out with friends, reading, etc. That's basically it.
Probably no international travel.
It will be like a simple retirement on a low income, because in a socialist system the resources must be rationed.
This will drive a lot of young ambitious people to insanity. Nothing meaningful for them to achieve. No purpose. Drug use, debauchery, depression, violence, degeneracy, gangs.
It will be a true idiocracy. No Darwinian selection pressures, unless the system enforces eugenics and population control.
Seems to me like our culture treats both survival and reproduction as an inalienable right. Most people would go so far as to say everyone deserves love, "there's a lid for every pot".
Maybe, if the only flavor of ambition you're aware of is that of SV types. Plenty of people have found achievement and meaning before and alongside the digital revolution world.
Jobs, careers, real work, all replaced by machines which can do it all better, faster, cheaper than humans.
Young people with modest ambitions to learn and master a skill and contribute to society, and have a meaningful life. That can be blue collar stuff too.
How will children respond to the question - "What do you want to be when you grow up?"
They can join the Amish communities where humans still do the work.
Yes, like retirement but without the old age. Right now I'm studying, so I do live on a very low income. But still, there are so many interesting things! For example, I'm trying to design a vacuum pump to 1mbar to be made of mostly 3d printed parts. Do vacuum pumps exist and can I buy them? Absolutely. But is it still fun to do the whole designing process? You bet. And I can't even start explaining all the things I'm learning.
> This will drive a lot of young ambitious people to insanity.
I teach teenagers in the age where they have to choose their profession. The ones going insane will be the unambitious people, those who just stay on TikTok all day and go to work because what else would they do? The ambitious will always have ideas and projects. And they won't mind creating something that already exists, just because they like the process of it.
We already see this with generative AI. Even though you could generate most of the images you'd want already, people still enjoy the process of painting or photographing. Humans are made to be creative and take pleasure from it, even if it is not economically valuable.
Hell, this is Hacker News. Hacking (in its original sense) was about creativity and problem-solving. Not because it will make you money, but because it was interesting and fun.
I am thinking about society as a whole, how it will affect all types of people and cultures on this planet.
> [...] how it will affect all types of people and cultures on this planet.
Some will definitely feel without purpose. But I'd argue that just having a job so that you have a purpose is just a band-aid, not a real solution. I won't say that purposelessness isn't a problem, just that it would be great to actually address the issue.
Granted, I do hold a utopic view. I continue to be curious due to my religious belief, where I'm looking forward to life unconstrained by age. Regardless whether this will manifest, I think it is healthy to remain curious and continue learning. So on "how it will affect all types of people": I really do think that people without purpose need to engage in curiosity and creativity, for their own mental health.
Introverts are only 25% - 40% of the population, and most people are not intellectually or artistically gifted (whether introvert or not), but they still want to contribute and feel valued by society.
> I'd argue that just having a job so that you have a purpose
It's not just about having a job. It's having an important or valuable role in society, feeling that your contributions actually matter to others - such as building or fixing things that others depend on, or providing for a family,
What would motivate a young boy to go through years of schooling, higher education, and so on, just to become a hobbyist, tinkering around on projects that no one else will ever use or really need? That may be acceptable for some niche personality types but not the majority.
Aspiring engineers or entrepreneurs are not merely motivated by having a job.
I am envisioning the AGI or ASI scenario which truly overtakes humans in all intellectual and physical capabilities, essentially making humans obsolete. That would smash the foundations and traditions of our civilization. It's an incredible gamble.
- Superintelligence poses an existential threat to humanity
- Predicting the future is famously difficult
- Given that uncertainty, we can't rule out the chance of our current AI approach leading to superintelligence
- Even a 1-in-1000 existential threat would be extremely serious. If an asteroid had a 1-in-1000 chance of hitting Earth and obliterating humanity we should make serious contingency plans.
Second question: how confident are you that you're correct? Are you 99.9% sure? Confident enough to gamble billions of lives on your beliefs? There are almost no statements about the future which I'd assign this level of confidence to.
I think the chance they're going to create a "superintelligence" is extremely small. That said I'm sure we're going to have a lot of useful intelligence. But nothing general or self-conscious or powerful enough to be threatening for many decades or even ever.
> Predicting the future is famously difficult
That's very true, but that fact unfortunately can never be used to motivate any particular action, because you can always say "what if the real threat comes from a different direction?"
We can come up with hundreds of doomsday scenarios, most don't involve AI. Acting to minimize the risk of every doomsday scenario (no matter how implausible) is doomsday scenario no. 153.
I'd say the chance that we never create a superintelligence is extremely small. You either have to believe that for some reason the human brain achieved the maximum intelligence possible, or that progress on AI will just stop for some reason.
Most forecasters on prediction markets are predicting AGI within a decade.
> that progress on AI will just stop for some reason
Yeah it might. I mean, I'm not blind and deaf, there's been tremendous progress in AI over the last decade, but there's a long way to go to anything superintelligent. If incremental improvement of the current state of the art won't bring superintelligence, can we be sure the fundamental discoveries required will ever be made? Sometimes important paradigm shifts and discoveries take a hundred years just because nobody made the right connection.
Is it certain that every mystery will be solved eventually?
There isn't an official precise definition of superintelligence, but it's usually vaguely defined as smarter than humans. Twice as smart would be sufficient by most definitions. We can be more conservative and say we'll only consider superintelligence achieved when it gets to 10x human intelligence. Under that conservative definition, 1/1000th of the performance of superintelligence would be 1% as smart as a human.
We don't have a great way to compare intelligences. ChatGPT already beats humans on several benchmarks. It does better than college students on college-level questions. One study found it gets higher grades on essays than college students. It's not as good as humans on long, complex reasoning tasks. Overall, I'd say it's smarter than a dumb human in most ways, and smarter than a smart human in a few ways.
I'm not certain we'll ever create superintelligence. I just don't see why you think the odds are "extremely small".
This extreme risk aversion and focus on negative outcomes is just the result of certain personality types, no amount of rationalizing will change your mind as you fundamentally fear the unknown.
How do you get out of bed everyday knowing there’s a chance you could get hit by a bus?
If your tribe invented fire you’d be the one arguing how we can’t use it for fear it might engulf the world. Yes, humans do risk starting wildfires, but it’s near impossible to argue the discovery of fire wasn’t a net good.
The new life form will be to humans, as humans are to chimps, or rats, or ants.
At this point we have lost control of the situation (the planet). We are no longer at the top of the food chain. Fingers crossed it all goes well.
It's an existential gamble. Is the gamble worth taking? No one knows.
You have cooked up a straw man that will believe anything as long as it contains a doomsday prediction. You are more than 99.9% confident about doomsday predictions, even if you claim you aren't.
I disagree at least on this one. I don't see any scenario where superintelligence comes into existence, but is for some reason limited to a mediocrity that puts it in contention with humans. That equilibrium is very narrow, and there's no good reason to believe machine-intelligence would settle there. It's a vanishingly low chance event. It considerably changes the later 1-in-n part of your comment.
So, since we've used the exact same reasoning to prove two opposite conclusions, it logically follows that this reasoning is faulty.
ASI to humans, is like humans to rats or ants.
Changing the premise to "superintelligence is the only thing that can save us" doesn’t invalidate the logic of being cautious. It just shifts the debate to which risk is more plausible. The reasoning about managing existential risks remains valid either way, the real question is which scenario is more likely, not whether the risk-based logic is flawed.
Just like with nuclear power, which can be both beneficial and dangerous, we need to be careful in how we develop and control powerful technologies. The recent deregulation by the US admin are an example of us doing the contrary currently.
Also, @tsimionescu's reasoning is spot on, and exactly how logic works.
Just like your proposition that any "small" chance justifies investing "everything" disregards the same argument regarding the precautionary principle of potentially devastating technologies. You've also slipped in an additonal "with no real downside" which you cannot predict with certainty anyways, rendering this argument infalsifiable. At least tsimionescu didn't dare making such a sweeping (but baseless) statement.
I'm guessing that you think that society is getting worse every year or will eventually collapse, and you hope that continued AI research might prevent that outcome.
I think you realise this is the weak point. You can't rule out the current AI approach leading to superintelligence. You also can't rule out a rotting banana skin in your bin spontaneously gaining sentience either. Does that mean you shouldn't risk throwing away that skin? It's so outrageous that you need at least some reason to rule it in. So it goes with current AI approaches.
What makes people think that the future advances in AI will continue to be linear instead of falling of and plateau? Don't all breakthrough technologies develop quickly at the start and then fall of in improvements as all the 'easy' improvements have already been made? In my opinion AI and AGI is like the car and the flying car. People saw continous improvements in cars and thought this rate of progress would continue indefinitely. Leading to cars that have the ability to not only drive but fly as well.
In the case of AGI we already know it is physically possible.
It was true before we allowed them to access external systems, disregarding certain rule which I forgot the origin.
The more general problem is a mix between the tradegy of the common; we have better understanding every passing day yet still don't understand exacly why LLM perform that well emergently instead of engineered that way; and future progress.
Do you think you can find a way around access boundaries to masquerade your Create/Update requests as Read in the log system monitoring it, when you have super intelligence?
LLMs are huge pretrained models. The economic benefit here is that you don't have to train your own text classification model anymore. (The LLM was likely already trained on whatever training set you could think of.)
That's a big time and effort saver, but no different from "AI" that we had decades prior. It's just more accessible to the normal person now.
Any of the signatories here match your criteria? https://safe.ai/work/statement-on-ai-risk#signatories
Or if you’re talking more about everyday engineers working in the field, I suspect the people soldering vacuum tubes to the ENIAC would not necessarily have been the same people with the clearest vision for the future of the computer.
Does the current AI give productivity benefits to writing code? Probably. Do OpenAI engineers have exclusive access to more capable models that give them a greater productivity boost than others? Also probably.
If one exclusive group gets the benefit of developing AI with a 20% productivity boost compared to others, and they develop a 2.0 that grants them a 25% boost, then a 3.0 with a 30% boost, etc...
The question eventually becomes, "is AGI technically possible"; is there anything special about meat that cannot be reproduced on silicon? We will find AGI someday, and more than likely that discovery will be aided by the current technologies. It's the path here that matters, not the specific iteration of generative LLM tech we happen to be sitting on in May 2025.
> If one exclusive group gets the benefit of developing AI with a 20% productivity boost compared to others, and they develop a 2.0 that grants them a 25% boost, then a 3.0 with a 30% boost, etc...
That’s a bit of a stretch, generative AI is least capable of helping with novel code such as needed to make AGI.
If anything I’d expect companies working on generative AI to be at a significant disadvantage when trying to make AGI because they’re trying to leverage what they are already working on. That’s fine for incremental improvement, but companies rarely ride one wave of technology to the forefront of the next. Analog > digital photography, ICE > EV, coal mining > oil, etc.
Then it looks like Company A spends 90% of time on novel research work (while LLMs do all the busy work) and Company B spends 5% of time on novel research work.
Just really think about what you just said, sure spend 5% of the time is on the bits nobody on earth has any idea how to accomplish that’s how people will approach this project. Organizationally the grunt work is a trivial rounding error vs the completely unbound we’ve got no idea how to solve this problems bits.
Look forward to re-living that shift from life-changing community resource to scammy and user-hostile
Then the thought came, when will they start showing ads here.
I like to think that if we learn to pay for it directly, or the open source models get good enough, we could still enjoy that simplicity and focus for quite a while. Here’s hoping!
These models being expensive leads me to think they will look at all methods of monetization possible when seeking profitability. Rather than ads being off the table, it could feasibly make ads be on the table sooner.
It's also why totalitarian regimes love it, they can simply train it to regurgitate a modified version of reality.
The $20 monthly payment is not enough though and companies like Google can keep giving away their AI for free till OpenAI is bankrupt.
Decades ago I worked for a classical music company, fresh out of school. "So.. how do you anticipate where the music trend is going", I once naively asked one of the senior people on the product side. "Oh, we don't. We tell people really quietly, and they listen". They and the marketing team spent a lot of time doing very subtle work, easily as much as anything big like actual advertisements. Things like small little conversations with music journalists, just a dropped sentence or two that might be repeated in an article, or marginally influence an article; that another journalist might see and have an opinion on, or spark some other curiosity. It only takes a small push and it tends to spread across the industry. It's not a fast process, but when the product team is capable of road-mapping for a year or so in advance, a marketing team can do a lot to prepare things so the audience is ready.
LLMs represent a scary capability to influence the entire world, in ways we're not equipped to handle.
replace LLMs with TV, or smartphones, or maybe even mcdonald's, and you've got the same idea. through TV, corporations got to control a lot of the social world and people's behavior.
Also the current crop of AI agents are just utter crap. But that's a skill issue of the people coding them, expect actual advances here soon.
That step, along with getting politicians to pass it, is the only thing that will stop that outcome.
Even if you take him at his word, incentives are hard to ignore (and advertising is a very powerful business model when your goal is to create something that reaches everyone)
1) The Pareto frontier of open LLMs will keep expanding. The breakneck pace of open research/development, combined with techniques like distillation will keep the best open LLMs pretty good, if not the best.
2) The cost of inference will keep going down as software and hardware are optimized. At the extreme, we're lookin toward bit-quantized LLMs that run in RAM itself.
These two factors should mean a good open LLM alternative should always exist, one without ulterior motives. Now, will people be able to have the hardware to run it? Or will users just put up with ads to use the best LLM? The latter is likely, but you do have a choice.
I have not seen anything from sama or pmarca that I would classify as “authoritarian”.
altman building a centralised authority of who will be classed as "human" is about as authoritarian as you could get
I doubt Worldcoin will actually manage to corner the market. But the point is, if it did, bad things would happen. Though, that’s probably true of most products.
>Liberalism is a political and moral philosophy based on the rights of the individual, liberty, consent of the governed, political equality, the right to private property, and equality before the law. Liberals espouse various and often mutually conflicting views depending on their understanding of these principles but generally support private property, market economies, individual rights (including civil rights and human rights), liberal democracy, secularism, rule of law, economic and political freedom, freedom of speech, freedom of the press, freedom of assembly, and freedom of religion. Liberalism is frequently cited as the dominant ideology of modern history.
Tim Apple did it too, and we don’t assume he’s an authoritarian now too, do we? I imagine they would probably have done similarly regardless of who won the election.
It sure seems like an endorsement, but I think it’s simply modern corporate strategy in the American regulatory environment, same as when foreign dignitaries stay in overpriced suites in the Trump hotel in DC.
Those who don’t kiss the ring are clearly and obviously punished. It’s not in the interest of your shareholders (or your launch partners) to be the tall poppy.
As for the shareholders, Cook was more than happy to "do the right thing" in the past, even when under pressure (https://en.wikipedia.org/wiki/Apple–FBI_encryption_dispute).
Furthermore, you are dead wrong on the last point. The “dispute” between the FBI and Apple is a fiction designed to restore public trust in Apple’s privacy stance following the Snowden revelations about FAA702 (aka PRISM) that shows that companies allow the USG warrantless access to their data in realtime via special APIs or portals.
https://www.reuters.com/article/us-apple-fbi-icloud-exclusiv...
The tech executives came to DC to meet with Obama in the wake of the whole Snowden thing to discuss it, though it was widely reported as being a consult on fixing healthcare.gov (lol) a few outlets reported it correctly. There are photos of the meeting kicking around.
I imagine the Apple-vs-the-FBI narrative (which is widely regarded as true and has resulted in mainstream false belief, such as yours demonstrated here) was borne directly out of these meetings.
Apple intentionally maintains access to the majority of their users’ data by the USG and the CCP (in their respective zones). It is required for them to continue operating in their current fashion. Every iMessage and (basically) every file in iCloud (photos included) is readable by Apple and the government. Apple has the technical capability to prevent this by migrating their userbase to e2ee systems, and they do not.
I firmly believe that this is by design, and that they would be very severely punished, legally or extralegally, if they changed the status quo.
You mean, AGI will benefit all of humanity like War on Terror spread democracy?
Regardless of intent, it was most definitely sold to the American public on that premise.
Google screamed against service revenue and advertising while building the world's largest advertising empire. Facebook screamed against misinformation and surveillance while enabling it on a global scale. Netflix screamed against the overpriced cable TV industry while turning streaming into modern overpriced cable television. Uber screamed against the entrenched taxi industry harming workers and passengers while creating an unregulated monster that harmed workers and passengers.
Altman and OpenAI are no different in this regard, loudly screaming against AI harming humanity while doing everything in their capacity to create AI tools that will knowingly harm humanity while enriching themselves.
If people trust the performance instead of the actions and their outcomes, then we can't convince them otherwise.
Condoning "honest liars" enables a whole other level of open and unrestricted criminality.
That doesn't mean it has to always be this way, though. Back when I had more trust in the present government and USPS, I mused on how much of a game changer it might be for the USPS to provide free hosting and e-mail to citizens, repurposing the glut of unused real estate into smaller edge compute providers. Everyone gets a web server and 5GB of storage, with 1A Protections letting them say and host whatever they like from their little Post Office Box. Everyone has an e-mail address tied to their real identity, with encryption and security for digital mail just like the law provides for physical mail. I still think the answer is about enabling more people to engage with the internet on their selective terms (including the option of disengagement), rather than the present psychological manipulation everyone engages in to keep us glued to our screens, tethered to our phones, and constantly uploading new data to advertisers and surveillance firms alike.
But the nostalgic view that the internet used to be different is just that: rose-tinted memories of a past that never really existed. The first step to fixing this mess is acknowledging its harm.
The Internet has changed a lot over the decades, and it did used to be different, with the differences depending on how many years you go back.
It absolutely did. Steve Wozniak was real. Silicon Valley wasn't always a hive of liars and sycophants.
We should acknowledge the past flatly and objectively for what it was and spend more time building that future, than listening to the victors of the past brag and boast, content to wallow in their accomplishments instead of rejoining contributors to tomorrow. The good leaders of yesteryear have stepped aside in lieu of championing newer, younger visionaries; those still demanding respect for what they did fifty years ago in circumstances we can only dream about, are part of the problem.
We should champion the good people who did the good things and managed to resist the temptations of the poisoned apple, but we shouldn’t hold an entire city on a pedestal because of nostalgia alone. Nobody, and no entity, is that deserving.
Nobody said there were no bastards. Just that they didn’t have dominion. We let this happen, in part by being lazy and cynical.
I think most people will snitch on bad behavior as children. However, our systems often allow other children to discipline the snitch, rather than correct the negative behavior the snitch raised. We see it in adult systems as well: whistleblowers often end up with substantially shorter and poorer lives for attempting to assert accountability or consequences on those who committed them, while the perpetrators often enjoy lives of immense wealth and reward regardless of the whistleblower's actions.
If you want people to stop being "lazy" and "cynical", then you have to support them when systems turn against them. In my experience, none of ya'll actually want to also walk out of work when layoffs happen following a profitable quarter for no other reason than to juice the share price, none of ya'll also want to walk off the job because your employer is taking contracts from authoritarian regimes, none of ya'll also want to put yourselves in the line of fire and risk harm over your purported values.
Don't blame us cynics when we have the battle scars showing our commitment to a better tomorrow. What have you done to prevent cynicism?
It was sparked by going to a video conference "Hyperlocal Heroes: Building Community Knowledge in the Digital Age" hosted by New_ Public: https://newpublic.org/ "Reimagine social media: We are researchers, engineers, designers, and community leaders working together to explore creating digital public spaces where people can thrive and connect."
A not-insignificant amount of time in that one-hour teleconference was spent related to funding models for local social media and local reporting.
Afterwards, I got to thinking. The USA spent literally trillions of dollars on the (so-many-problematical-things-about-it-I-better-stop-now) Iraq war. https://en.wikipedia.org/wiki/Financial_cost_of_the_Iraq_War "According to a Congressional Budget Office (CBO) report published in October 2007, the US wars in Iraq and Afghanistan could cost taxpayers a total of $2.4 trillion by 2017 including interest."
Or, from a different direction, the USA spends about US$200 billion per year on mostly-billboard-free roads: https://www.urban.org/policy-centers/cross-center-initiative... "In 2021, state and local governments provided three-quarters of highway and road funding ($154 billion) and federal transfers accounted for $52 billion (25 percent)."
That's about US$700 per person per year on US roads.
So, clearly huge amounts of money are available in the USA if enough people think something is important. Imagine if a similar amount of money went to funding exactly what you outlined -- a free web presence for distributed social media -- with an infrastructure funded by tax dollars instead of advertisements. Isn't a healthy social media system essential to 21st century online democracy with public town squares?
And frankly such a distributed social media ecosystem in the USA might be possible for at most a tenth of what roads cost, like perhaps US$70 per person per year (or US$20 billion per year)?
Yes, there are all sorts of privacy and free speech issues to work through -- but it is not like we don't have those all now with the advertiser-funded social media systems we have. So, it is not clear to me that such a system would be immensely worse than what we have.
But what do I know? :-) Here was a previous big government suggestion be me from 2010 -- also mostly ignored (until now 15 years later the USA is in political crisis over supply chain dependency and still isn't doing anything very related to it yet): "Build 21000 flexible fabrication facilities across the USA" https://web.archive.org/web/20100708160738/http://pcast.idea... "Being able to make things is an important part of prosperity, but that capability (and related confidence) has been slipping away in the USA. The USA needs more large neighborhood shops with a lot of flexible machine tools. The US government should fund the construction of 21,000 flexible fabrication facilities across the USA at a cost of US$50 billion, places where any American can go to learn about and use CNC equipment like mills and lathes and a variety of other advanced tools and processes including biotech ones. That is one for every town and county in the USA. These shops might be seen as public extensions of local schools, essentially turning the shops of public schools into more like a public library of tools. This project is essential to US national security, to provide a technologically literate populace who has learned about post-scarcity technology in a hands-on way. The greatest challenge our society faces right now is post-scarcity technology (like robots, AI, nanotech, biotech, etc.) in the hands of people still obsessed with fighting over scarcity (whether in big organizations or in small groups). This project would help educate our entire society about the potential of these technologies to produce abundance for all."
When we already have efficient food production that drove down costs and increased profits (a good thing), what else is there for companies to optimize for, if not loading it with sugar, putting it in cheap plastic, bamboozling us with ads?
This same dynamic plays out in every industry. Markets are a great thing when the low hanging fruit hasn't been picked, because the low hanging fruit is usually "cut the waste, develop basic tech, be efficient". But eventually the low hanging fruit becomes "game human's primitive reward circuits".
It’s just that their biases have much more capacity to cause damage as their wealth gives them so much power.
> They are roughly as delusional as everyone else.
I would bet serious money that people who believe in Ayn Rand are generally more delusional than others, and the same goes for the ultra-wealthy living in a bubble of sycophants.
And their wealth gives them much more capacity - and motive - to cause damage.
But once you control a significant enough chunk of money, it becomes clear the pie doesn't get any bigger the more shiny coins you have, you only have more relative purchasing power, automatically making everyone else poorer.
Altman keeps on talking about AGI as if we're already there.
But reasonable people could argue that we've achieved AGI (not artificial super intelligence)
https://marginalrevolution.com/marginalrevolution/2025/04/o3...
Fwiw, Sam Altman will have already seen the next models they're planning to release
What it really says is that if a user wants to control the interaction and get the useful responses, direct programmatic calls to the API that control the system prompt are going to be needed. And who knows how much longer even that will be allowed? As ChatGPT reports,
> "OpenAI has updated the ChatGPT UI (especially in GPT-4-turbo and ChatGPT Plus environments) to no longer expose the full system prompt or baseline prompt directly."
Can some business person give us a summary on PBCs vs. alternative registrations?
(IANAL but run a PBC that uses this charter[1] and have written about it here[2] as part of our biennial reporting process.)
[1] https://github.com/OpenCoreVentures/ocv-public-benefit-compa...
[2] https://goauthentik.io/blog/2024-09-25-our-biennial-pbc-repo...
Theory: It allows the CEO to make decisions motivated not just by maximizing shareholder value but by some other social good. Of course, very few PBC CEOs choose to do that.
I personally think the conversation, including obviously in the post itself, has swung too far in the direction of how AGI can or will potentially affect the ethical landscape regarding AI, however. I think we really ought to concern ourselves with addressing and mitigating effects that it already HAS brought - both good and bad - rather than engaging in any excessive speculation.
That's just me, though.
If the entrenched giants (Google, Microsoft and Apple) catch up - and Google 100% has, if not surpassed - they have a thousand levers to pull and OpenAI is done for. Microsoft has realized this, hence why they're breaking up with them - Google and Anthropic have shown they don't need OpenAI. Galaxy phones will get a Gemini button, Chrome will get it built into the browser. MS can either develop their own thing , use opensource models, or just ask every frontier model provider (and there's already 3-4 as we speak) how cheaply they're willing to deliver. Then chuck it right in the OS and Office first-class. Which half the white collar world spends their entire day staring at. Apple devices too will get an AI button (or gesture, given it's Apple) and just like MS they'll do it inhouse or have the providers bid against each other.
The only way OpenAI David was ever going to beat the Goliaths GMA in the long run was if it were near-impossible to catch up to them, á la TSMC/ASML. But they did catch up.
OpenAI has been on a winning streak that makes ChatGPT the default chatbot for most of the planet.
Everybody else like you describe is trying to add some AI crap behind a button on a congested UI.
B2B market will stay open but OpenAI has certainly not peaked yet.
What network effect does OpenAI have? Far as I can tell, moving from OpenAI to Gemini or something else is easy. It’s not sticky at all. There’s no “my friends are primarily using OpenAI so I am too” or anything like that.
So again, I ask, what makes it sticky?
At best they have a bit of cheap tribalism that might prevent some incurious people who don't care much about using the best tools noticing that they aren't.
So the interesting question is: How did that happen? Why wasn't Google search an easily swapped commodity? Or if it was, how did they win and defend their default status? Why didn't the existing juggernauts at the time (Microsoft) beat them at this game?
I have my own answers for these, and I'm sure all the smart people figuring out strategy at Open AI have thought about similar things.
It's not clear if Open AI will be able to overcome this commodification issue (personally, I think they won't), but I don't think it's impossible, and there is prior art for at least some of the pages in this playbook.
Google is doing well for the moment, but OpenAI just closed a $40 billion round. Neither will be able to rest for a while.
Maybe the big amount of money they've given to Apple which is their direct competitor in the mobile space. Also good amount of money given to Firefox, which is their direct competitor in the browser space, alongside side Safari from Apple.
Most people don't care about the search engine. The default is what they will used unless said default is bad.
So then apply that to Open AI. What are the distribution channels? Should they be paying Cursor to make them the default model? Or who else? Would that work? If not, why not? What's different?
My intuition is that this wouldn't work for them. I think if this "pay to be default" strategy works for someone, it will be one of their deeper pocketed rivals.
But I also don't think this was the only reason Google won search. In my memory, those deals to pay to be the default came fairly long after they had successfully built the brand image as the best search engine. That's how they had the cash to afford to pay for this.
A couple years ago, I thought it seemed likely that Open AI would win the market in that way, by being known as the clear best model. But that seems pretty unclear now! There are a few different models that are pretty similarly capable at this point.
Essentially, I think the reason Google was able to win search whereas the prospects look less obvious for Open AI is that they just have stronger competition!
To me, it just highlights the extent to which the big players at the time of Google's rise - Microsoft, Yahoo, ... Oracle maybe? - really dropped the ball on putting up strong competition. (Or conversely, Google was just further ahead of its time.)
They have the brand recognition and consumer goodwill no other brand in AI has, incredibly so with school students, who will soon go into the professional world and bring that goodwill with them.
I think better models are enough to dethrone OpenAI in API, B2C and internal enterprise use cases, but OpenAI has consumer mindshare, and they're going to be the king of chatbots forever. Unless somebody else figures out something which is better by orders of magnitude and that Open AI can't copy quickly, it's going to stay that way.
Apple had the opportunity to do something really great here. With Siri's deep device integration on one hand and Apple's willingness to force 3rd-party devs to do the right thing for users on the other, they could have had a compelling product that nobody else could copy, but it seems like they're not willing to go that route, mostly for privacy, antitrust and internal competency reasons, in that order. Google is on the right track and might get something similar (although not as polished as typical Apple) done, but Android's mindshare among tech-savvy consumers isn't great enough for it to get traction.
...Until their employer forces them to use Microsoft Copilot, or Google Gemini, or whatever, because that's what they pay for and what integrates into their enterprise stack. And the new employee shrugs and accepts it.
...yes. Office is the market leader. Slack has between a fifth and a fourth of the market. Coca-Cola's products have like 70% market share in the American carbonated soft-drink market [1].
[1] https://www.investopedia.com/ask/answers/060415/how-much-glo...
My friend teaches at a Catholic girls’ high school and based on what he tells me, everyone knows about ChatGPT, both staff and students. He just had to fail an entire class on an assignment because they all used it to write a book summary (which many of them royally screwed up because there’s another book with a nearly identical title).
It’s all anecdotal and whatnot but I don’t think many of them even know about Claude or Gemini, while ChatGPT has broad adoption within education. (I’m far less clear on how much mindshare it has within the general population though)
This will happen, and it won't be another model which Open AI can't copy, it'll be products.
I don't doubt OpenA I can create the better models but they're no moat if they're not in better products. Right now the main product is chat, which is easy enough to build, but as integrations get deeper how can OpenAI actually ensure it keeps traffic?
Case in point, Siri. Apple allows you to use ChatGPT with Siri right now. If Apple chooses so, they could easily remove that setting. On most devices ChatGPT lives within the confines of an app or the browser. A phone with deep AI integration is arguably a fantastic product— much better than having to open an app and chat with a model. How quickly could Open AI build a phone that's as good as those of the big phone companies today?
To draw a parallel— Google Assistant has long been better than Siri, but to use Siri you don't have to install an app. I've used both Android and iOS, and every time I'm on iPhone I switch back to Siri because in spite of being a worse assistant, it's overall a better product. It integrates well with the rest of the phone, because Apple has chosen to not allow any other voice assistant integrate deeply with the rest of the phone.
If you look at Gemini, I know people using it daily.
And nobody's saying OpenAI will go bankrupt, they'll certainly continue to be a huge player in this space. But their astronomical valuation was based on the initial impression that they were the only game in town, and it will come down now that that's no longer true. Hence why Altman wants to cash out ASAP.
My impression is that Claude is a lot more popular – and it’s the one I use myself, though as someone else said the vast majority of people, even in software engineering, don’t use AI often at all.
Facebook wasn't some startup when Google+ entered the scene; they were already cash flow positive, and had roughly 30% ads market share.
OpenAI is still operating at a loss despite having 50+% of the chatbot "market". There is no easy path to victory for them here.
This moat is non-existent when it comes to Open AI.
All dissidents went into Little Wadyia.
When the Dictator himself visited it, he started to fake his name by copying the signs and names he saw on the walls. Everyone knew what he was.
Internet social networks are like that.
Now, this moat thing. That's hilarious.
For example, I'd never suggest that e.g. MS could take on TikTok, despite all the levers they can pull, and being worth magnitudes more. No chance.
Consumer brand companies such as Coca Cola and Pepsi spend millions on brand awareness advertising just to be the “default” in everyone’s heads. When there’s not much consequence choosing one option over another, the one you’ve heard of is all that matters
OpenAI has like 10 to 20% market share [1][2]. They're also an American company whose CEO got on stage with an increasingly-hated world leader. There is no universe in which they keep equal access to the world's largest economies.
[1] https://iot-analytics.com/leading-generative-ai-companies/
[2] https://www.enterpriseappstoday.com/stats/openai-statistics....
LLMs themselves aren't the moat, product integration is. Google, Apple and Microsoft already have the huge user bases and platforms with a big surface area covering a good chunk of our daily life, that's why I think they're better positioned if models become a commodity. OpenAI has the lead now, but distribution is way more powerful in the long run.
There is little to no money to be made in GAI, it will never turn into AGI, and people like Altman know this, so now they’re looking for a greater fool before it is too late.
In this niche you can be irrellevant in months when your models drop behind.
The news that they did that would make them lose most of their revenue pretty fast.
OpenAI has claimed this. But Altman is a pathological liar. There are lots of ways of disguising operating costs as capital costs or R&D.
OpenAI models are already of the most expensive, they don’t have a lot of levers to pull.
I feel like people overuse this criticism. That's not the only way that companies with a lot of revenue lose money. And this isn't at all what OpenAI is doing, at least from their customers' perspective. It's not like customers are subscribing to ChatGPT simply because it gives them something they were going to buy anyway for cheaper.
It’s ok to not buy into the vision or think it’s impossible. But it’s a shallow dismissal to make the unnuanced comparison, especially when we’re talking about a brand new technology - who knows what the cost optimization levers are. Who knows what the market will bear after a few more revs.
When the iPhone first came out, it was too expensive, didn’t do enough, and many people thought it was a waste of apples time when they should be making music players.
This comparison is always used when people are trying to hype something. For every "iPhone" there are thousands of failures
> But it’s a shallow dismissal to make the unnuanced comparison, especially when we’re talking about a brand new technology - who knows what the cost optimization levers are. Who knows what the market will bear after a few more revs.
You're acting as-if OpenAI is still the only player in this space. OpenAI has plenty of competitors who can deliver similar models for cheaper. Gemini 2.5 is an excellent and affordable model and Google has a substantially better capacity to scale because of a multi-year investment in its TPUs.
Whatever first mover advantage OpenAI had has been quickly eliminated, they've lost a lot of their talent, and the chief hypothesis they used to attract the capital they've raised so far is utterly wrong. VCs would be mad to be continuing to pump money into OpenAI just to extend their runway -- at 5 Bln losses per year they need to actually consider cost, especially when their frontier releases are only marginal improvements over competitors.
... this is a bubble despite the promise of the technology and anyone paying attention can see it. For all of the dumb money employed in this space to make it out alive, we'll have to at least see a fairly strong form of AGI developed, and by that point the tech will be threatening the general economic stability of the US consumer.
Is that a bubble? I suppose it is; it’s also probably the right strategy.
Why does the forum of an incubator that now has a portfolio that is like 80% AI so routinely bearish on AI? Is it a fear of irrelevance?
When the Internet was developed they didn't imagine the world wide Web.
When cars started to get popular people still thought there would be those who are going to stick with horses.
I think you're right on the AI we're just on the cusp of it and it'll be a hundred times bigger than we can imagine.
Back when oil was discovered and started to be used it was about equal to 500 laborers now automated. One AI computer with some video cards are now worth x number of knowledge workers. That never stop working as long as the electricity keeps flowing.
Even five years into this "AI revolution," the boosters haven't been able to paint a coherent picture of what AI could reasonably deliver – and they've delivered even less.
The world is changing and that is scary.
This makes me want to invest in malpractice lawyers, not OpenAI
The other point is still suspect: that LLMs will ever scale to AGI.
Which specifically means reliability and explainability for higher-order thinking.
The writing is on the wall that LLMs are going to automate failure-tolerant work.
But the rub there is that failure-tolerant work is also tolerant of less than state of the art, cost-optimized LLMs.
Which leaves OpenAI where? AGI or bust.
And I wouldn't take that bet, when MS, Google, and Apple are alternative options.
I don't think there is serious argument that LLMs won't generate tremendous value. The question is who will capture it. PCs generated massive value. But other than a handful of manufacturers and designers (namely, Apple, HP, Lenovo, Dell and ASUS), most PC builders went bankrupt. And out of the value generated by PCs in the world, the vast majority was captured by other businesses and consumers.
Oh we know: https://pmc.ncbi.nlm.nih.gov/articles/PMC11006786/
The article could just as easily be about “Delayed diagnosis of a transient ischemic attack caused by talking to some rando on Reddit” and it would be just as (non) newsworthy.
AI isn't going to be the world changing, AGI, that was sold to the public. Instead, it will simply be another B2B SaaS product. Useful, for sure. Even profitable for startups.
But "take over the world" good? Unlikely.
The fact that people know Coca Cola doesnt mean they drink it.
That name recognition made Coca Cola into a very successful global corporation.
Knowing and using are not the same thing.
40% of the US is a huge customer base.
The names don't even matter when everything is baked in.
Slack? Zoom? Teams?
I'm sure you'd get a somewhat uniform distribution.
Ask the same today, and I'd bet most will say Teams. Why Teams? Because it comes with office / windows, so that's what most people will use.
Same logic goes for the AI / language models...which one are people going to use? The ones that are provided as "batteries included" in whatever software or platform they use the most. And for the vast majority of regular people / workers, it is going to be something by microsoft / google / whatever.
1: https://www.techpolicy.press/transcript-senate-judiciary-sub...
The only thing OpenAI has right now is the ChatGPT name, which has become THE word for modern LLMs among lay people.
Anecdotally, I've switched to Gemini as my daily driver for complex coding tasks. I prefer Claude's cleaner code, but it is less capable at difficult problems, and Anthropic's servers are unreliable.
The wisest move in the chatbot business might be to wait and see if anyone discovers anything profitable before spending more effort and wasting more money on chat R&D, which includes most agentic stuff. Reliable assistants or something along those lines might be the next big breakthrough (if you ask certain futurologists), but the technology we have seems unsuitable for any provable reliability.
ML can be applied in a thousand ways other than LLMs, and many will positively impact our lives and create their own markets. But OpenAI is not in that business. I think the writing is on the wall, and Sama's vocal fry, "AGI is close," and humanity verification crypto coins are smoke and mirrors.
Most people in society connect AI directly to ChatGPT and hence OpenAI. And there has been a lot of progress in image generation, video generation, ...
So I think your timeline and views are slightly off.
GPT-2 was released in 2019, GPT-3 in 2020. I'd say 2020 is significant because that's when people seriously considered the Turing test passed reliably for the first time. But for the sake of this argument, it hardly matters what date years back we choose. There's been enough time since then to see the plateau.
> Most people in society connect AI directly to ChatGPT and hence OpenAI.
I'd double-check that assumption. Many people I've spoken to take a moment to remember that "AI" stands for artificial intelligence. Outside of tongue-in-cheek jokes, OpenAI has about 50% market share in LLMs, but you can't forget that Samsung makes AI washing machines, let alone all the purely fraudulent uses of the "AI" label.
> And there has been a lot of progress in image generation, video generation, ...
These are entirely different architectures from LLM/chat though. But you're right that OpenAI does that, too. When I said that they don't stray much from chat, I was thinking more about AlexNet and the broad applications of ML in general. But you're right, OpenAI also did/does diffusion, GANs, transformer vision.
This doesn't change my views much on chat being "not seeing the forest for the trees" though. In the big picture, I think there aren't many hockey sticks/exponentials left in LLMs to discover. That is not true about other AI/ML.
This site and many others were littered with OpenAI stories calling it the next Bell Labs or Xerox PARC and other such nonsense going back to 2016.
And GPT stories kicked into high gear all over the web and TV in 2019 in the lead-up to GPT-2 when OpenAI was telling the world it was too dangerous to release.
Certainly by 2021 and early 2022, LLM AI was being reported on all over the place.
>For most of the world LLM's did not exist before those dates.
Just because people don't use something doesn't mean they don't know about it. Plenty of people were hearing about the existential threat of (LLM) AI long before ChatGPT. Fox News and CNN had stories on GPT-2 years before ChatGPT was even a thing. Exposure doesn't get much more mainstream than that.
No, it wasn't.
As a proxy, here's HN results prior to November, 2022 - 13 results.
https://hn.algolia.com/?dateEnd=1667260800&dateRange=custom&...
Here's Google Trends, showing a clear uptick May 2023, and basically no search volume before (the small increase Feb. 2023 probably Meta's Llama).
https://trends.google.com/trends/explore?date=today%205-y&ge...
https://trends.google.com/trends/explore?date=today%205-y&ge...
As another proxy, compare Nvidia revenues - $26.91bln in 2022, $26.97bln in 2023, $60bln 2024, $130bln 2025. I think it's clear the hype didn't start until 2023.
You're welcome to point out articles and stores before this time period "hyping" LLM's, but what I remember is that before ChatGPT there was very little conversation around LLM's.
Image generation was also a continuous slope of hype all the way from the original GAN, then thispersondoesnotexist, the sketch-to-photo toys by Nvidia and others, the avocado sofa of DallE. Then DallE2, etc.
The hype can continue to grow beyond our limit of perception. For people who follow such news their hype sensor can be maxed out earlier, and they don't see how ridiculously broadly it has spread in society now, because they didn't notice how niche it was before, even though it seemed to be "everywhere".
We do appear to be hitting a cap on the current generation of auto-regressive LLMs, but this isn't a surprise to anyone on the frontier. The leaked conversations between Ilya, Sam and Elon from the early OpenAI days acknowledge they didn't have a clue as to architecture, only that scale was the key to making experiments even possible. No one expected this generation of LLMs to make it nearly this far. There's a general feeling of "quiet before the storm" in the industry, in anticipation of an architecture/training breakthrough, with a focus on more agentic, RL-centric training methods. But it's going to take a while for anyone to prove out an architecture sufficiently, train it at scale to be competitive with SOTA LLMs and perform enough post training, validation and red-teamint to be comfortable releasing to the public.
Current LLMs are years and hundreds of millions of dollars of training in. That's a very high bar for a new architecture, even if it significantly improves on LLMs.
"It just suddenly appeared out of nowhere" is just a perception based on missing info. Many average people think ChatGPT was a sudden innovation specifically by OpenAI seemingly out of nowhere. Because they didn't follow it.
The more you know about it, the less groundbreaking it is.
Personally, deep research and o3 have been transformative, taking LLMs from something I have never used to something that I am using daily.
Even if the progress ends up plateauing (which I do not believe will happen in the near term), behaviors are changing; OpenAI is capturing users, and taking them from companies like Google. Google may be able to fight back and win - Gemini 2.5 Pro is great - but any company sitting this out risks being unable to capture users back from Open AI at a later date.
> taken it from a toy to genuinely insanely useful.
Really?
Why? I paid for Claude for a while, but with Deepseek, Gemini and the free hits on Mistral, ChatGPT, Claude and Perplexity I'm not sure why I would now. This is anecdotal of course, but I'm very rarely unique in my behaviour. I think the best the subscription companies can hope for is that their subscribers don't realize that Deepseek and Gemini can basically do all you need for free.
I cannot stress this enough: if you know what Deepseek, Claude, Mistral, and Perplexity are, you are not a typical consumer.
Arguably, if you have used a single one of those brands you are not a typical consumer.
The vast majority of people have used ChatGPT and nothing else, except maybe clicking on Gemini or Meta AI by accident.
They might not “know” the brand as well as ChatGPT, but the average consumer has definitely been exposed to those at the very least.
DeepSeek also made a lot of noise, to the point that, anecdotally, I’ve seen a lot of people outside of tech using it.
Why are you still pretending anything is going to come out of this?
If every major player had an AI option, i'm just not understanding how because OpenAi moved first or got big first, the hugely massively successful companies that did the same thing for multiple decades don't have the same advantage?
1. OpenAI is apparently in the process of building a social network.
2. OpenAI is apparently working with Jonny Ive on some sort of hardware.
3. OpenAI is increasingly working on "memory" as a LLM feature. Users may be less likely to switch as an LLM increasingly feels like a person that knows you, understands you, has a history with you, etc.
4. Google and MSFT are leveraging their existing strengths. Perhaps you will stick with Gemini given deep integration with Android, Google Drive, Sheets, Docs, etc.
5. LLMs, as depressing as this sounds, will increasingly be used for romantic/friend purposes. These users may not want to switch, as it would be like breaking up and finding a new partner.
6. Your chat history, if it can't be easily exported/imported, may be a sticky feature, especially if it can be improved (e.g. easily search, cross-reference, chats, like a supercharged interconnecting note app with brains).
I could list 100 more of these. Perhaps none of the above will happen, but again, they have 400M weekly users and they will find ways to keep them. It's a lot easier to keep users that have a habit of showing up, then getting them in the first place. There's a reason that Google is treating this like an emergency; they are at serious risk of having their search cash cow permanently disrupted if they don't act fast to win.
#5 stands out as well as a substantial barrier.
The rest to me our sticky, but no more uniquely sticky than any other service that retains data. Like the switching cost of email or a browser. It does stick but not insurmountable and once the switch is made, it's like why did I wait so long? (I'm a Safari user!)
Anyway, thanks for the thoughtful reply.
[1] https://www.reforge.com/podcast/unsolicited-feedback/the-gre...
Google is alright, but they have similar stupid noncompete vendor lock in rule, and no way to opt out of training, so there’s no real reason to trust Google. Yeah they could ship tool use in reasoning to catch up to o3, but it’ll just be catching up and not passing unless they fix the stupid legal terms.
Claude IDK how to trust, they train on feedback and everything is feedback, and they have the noncompete rule written even more broadly, dumb to use that.
Grok has a noncompete rule but also has a way to opt out of training, so it’s on the same tier of ClosedAI. I use it sometimes for jokey toy image generation crap but there’s no way to use it for anything serious since it has a copypasted closed ai prohibition
Mistral needs better models and simpler legalese, it’s so complicated and impossible to know which of the million legal contracts applies
IMHO meta is the only player, but they shot themselves in the foot by making Llama 4 too big for the local llama community to even use, super dumb, killed their most valuable thing which was the community.
That means the best models we can use for work without needing to worry about a lawsuit, are Qwen, and DeepSeek distills, no American AI is even in the same ballpark, Gemma 3 is refusal king if you even hint at something controversial. basically, America is getting actively stomped by China in AI right now, because their stuff is open and interoperable, and ours is closed and has legal noncompete bullshit, what can we actually build that doesn’t compete with these companies? Nothing
I'd say Chain-of-Thought has massively improved LLM output. Is that "incremental"? Why is that more incremental than the move from GPT-2 to GPT-3? Sure you can say that this is when LLMs first passed some sort of Turing test, but fundamentally there was no technological difference from GPT-3 to GPT-4. In fact I would say the quality of GPT-4 unlocked thousands (millions?) more use-cases that were not very viable with the quality delivered by GPT-3. I don't see any reason for more use-cases to keep being unlocked by LLM improvements.
Market share of OpenAI is like 90%+.
Source? I've seen 10 to 20% [1][2].
[1] https://iot-analytics.com/leading-generative-ai-companies/
[2] https://www.enterpriseappstoday.com/stats/openai-statistics....
I probably need to clarify what I'm talking about, so that peeps like @JumpCrisscross can get a better grasp of it.
I do not mean the total market share of the category of businesses that could be labeled as "AI companies", like Microsoft or NVIDIA, on your first link.
I will not talk about your second link because it does not seem to make sense within the context of this conversation (zero mentions or references to market share).
What I mean is:
* The main product that OpenAI sells is AI models (GPT-4o, etc...)
* OpenAI does not make hardware. OpenAI is not in the business of cloud infrastructure. OpenAI is not in the business of selling smartphones. A comparison between OpenAI and any of those companies would only make sense for someone with a very casual understanding of this topic. I can think of someone, perhaps, who only used ChatGPT a couple times and inferred it was made by Apple because it was there on its phone. This discussion calls for a deeper understanding of what OpenAI is.
* Other examples of companies that sell their own AI models, and thus compete directly with OpenAI in the same market that OpenAI operates by taking a look at their products and services, are Anthropic (w/ Claude), Google (w/ Gemini) and some others ones like Meta and Mistral with open models.
* All those companies/models, together, make up some market that you can put any name you want to it (The AI Model Market TM)
That is the market I'm talking about, and that is the one that I estimated to be 90%+ which was pretty much on point, as usual :).
1: https://gs.statcounter.com/ai-chatbot-market-share
2: https://www.ctol.digital/news/latest-llm-market-share-mar-20...
Your second source doesn’t say what it’s measuring and disclaims itself as from its “‘experimental era’ — a beautiful mess of enthusiasm, caffeine, and user-submitted chaos.” Your first link only measures chatbots.
ChatGPT is a chatbot. OpenAI sells AI models, including via ChatGPT. Among chatbots, sure, 84% per your source. (Not “90%+,” as you stated.) But OpenAI makes more than chatbots, and in the broader AI model market, its lead is far from 80+ percent.
TL; DR It is entirely wrong to say the “market share of OpenAI is like 90%+.”
[1] https://firstpagesage.com/reports/top-generative-ai-chatbots...
>10%-20%
Lmao, not even in Puchal wildest dreams.
One, you suggested OP had not “looked at the actual numbers.” That implies you have. If you were just guessing, that’s misleading.
Two, you misquoted (and perhaps misunderstand) a statistic that doesn’t match your claim. Even in your last comment, you defined the market as “companies that sell their own AI models” before doubling down on the chatbot-only figure.
> not even in Puchal wildest dreams
Okay, so what’s your source? Because so far you’ve put forward two sources, a retracted one and one that measures a single product that you went ahead and misquoted.
OpenAI trained GPT-4.1 and 4.5—both originally intended to be GPT-5 but they were considered disappointments, which is why they were named differently. Did they really believe that scaling the number of parameters would continue indefinitely without diminishing returns? Not only is there no moat, but there's also no reasonable path forward with this architecture for an actual breakthrough.
No more caps on profit, a simpler structure to sell to investors, and Altman can finally get that 7% equity stake he's been eyeing. Not a bad outcome for him given the constraints apparently imposed on them by "the Attorney General of Delaware and the Attorney General of California".
Let's see how this plays out. PBC effectively means nothing - just take a look at Xai and its purchase of Twitter. I would love to listen reasoning explaining why this ~33 billion USD move is benefiting public.
Right; so, "Worker Unions" work.
edit: to be clear, it's not a bad thing - we should want companies that create consumer surplus. But that's the default state of companies in a healthy market.
This is true for literally any transaction. Actually, it's true for any rational action. If you're being tortured, and you decide it's not worth it to keep your secrets hidden any longer, you get more than you give up when you stop being tortured.
It’s called prisoners dilemma when even the government is propping this up.
Not being snarky here, like what is the purported thesis behind them?
Some founders truly believe in structuring the company for the benefit of the public, but Altman has already shown he's not one of them.
There was never a coherent explanation of its firing the CEO.
But they could have stuck with that decision if they believed in it.
Then things went unexpectedly well, people were valuing them at billions of dollars, and they suddenly decided they weren't open any more. Suddenly they were all about Altman's Interests Safety (AI Safety for short).
The board tried to fulfil its obligation to get the nonprofit to do the things in its charter, and they were unsuccessful.
But they found themselves alone in that it turns out the employees (who were employed by the for-profit company) and investors (MSFT in particular) didn't care about the mission and wanted to follow the money instead.
So the board had no choice but to capitulate and leave.
Being rich results in a kind of limitation of scope for ambition. To the sufferer, a person who has everything they could want, there is no other objective worth having. They become eccentric and they pursue more money.
We should have enrichment facilities for these people where they play incremental games and don’t ruin the world like the paperclip maximizers they are.
The dude announces new initiatives from the White House, regularly briefs Senators and senior DoD leaders, and is the top get for interviews around the world for AI topics.
There’s a lot more to be ambitious about than just money.
Maybe he wants to use the money in some nebulous future way, subjugating all people in a way that deals with his childhood trauma or whatever. That’s also something rich people do when they need a hobby aside from gathering more money. It’s not their main goal, except when they run into setbacks.
People are not complicated when they are money hoarders. They might have had hidden depths once, but they are thin furrows in the ground next to the giant piles of money that define them now.
So he doesn't enjoy the attention? Prestige or power? Respect?
Are you Sam Altman? Because you're making a lot of assumptions on his psyche right now.
Google/Anthropic are catching up, or already surpassed.
--Gordon Gekko
St. Altman plans to create a corporate god for us dumb schmucks, and he will be it's prophet.
This is already impossibly hard. Approximately zero people commenting would be able to win this battle in Sam’s shoes. What would they need to do to begin to have a chance? Rather than make all the obvious comments “bad evil man wants to get rich”, think what it would take to achieve the mission. What would you need to do in his shoes, aside from just give up and close up shop? Probably this, at the very least.
Edit: I don’t know the guy and many near YC do. So I accept there may be a lens I don’t have. But I’d rather discuss the problem, not the person.
> Sam’s Letter to Employees.
> OpenAI is not a normal company and never will be.
Where did I hear something like that before...
> Founders' IPO Letter
> Google is not a conventional company. We do not intend to become one.
I wonder if it's intentional or perhaps some AI-assisted regurgitation prompted by "write me a successful letter to introduce a new corporate structure of a tech company".
OpenAI admitting that they're not going to win?
> A lot of people around OpenAI in the early days thought AI should only be in the hands of a few trusted people who could “handle it”.
Whether they are a net positive or a net negative is arguable. If it's a net negative, then unleashing them to the masses was maybe the danger itself.
There is a lot to criticize about OpenAI and Sama, but this isn't it.
* The nonprofit is staying the same, and will continue to control the for-profit entity OpenAI created to raise capital
* The for-profit is changing from a capped-profit LLC to a PBC like Anthropic and Xai
* These changes have been at least tacitly agreed to by the attorneys general of California and Delaware
* The non-profit won’t be the largest shareholder in the PBC (likely Microsoft) but will retain control (super voting shares?)
* OpenAI thinks there will be multiple labs that achieve AGI, although possibly on different timelines
It's just that this bait has a shelf life and it looks like it's going to expire soon.
They already fight transparency in this space to prevent harmful bias. Why should I believe anything else they have to say if they refuse to take even small steps toward transparency and open auditing?
And the investors wailed and gnashed their teeth but it’s true, that is what they agreed to, and they had no legal recourse. And OpenAI’s new CEO, and its nonprofit board, cut them a check for their capped return and said “bye” and went back to running OpenAI for the benefit of humanity. It turned out that a benign, carefully governed artificial superintelligence is really good for humanity, and OpenAI quickly solved all of humanity’s problems and ushered in an age of peace and abundance in which nobody wanted for anything or needed any Microsoft products. And capitalism came to an end.
https://www.bloomberg.com/opinion/articles/2023-11-20/who-co...
We know it's a sword. And there's war, yadda yadda. However, let's do the cultivating thing instead.
What other AI players we need to convince?
sam altman: "OpenAI is not a normal company and never will be."
Hmmm
More crucially, since OpenAI's founding and especially over the past 18 months, it's grown increasingly clear that AI leadership probably won't be dominated by one company, progress of "frontier models" is stalling while costs are spiraling, and 'Foom' AGI scenarios are highly unlikely anytime soon. It looks like this is going to be a much longer, slower slog than some hoped and others feared.
Then why is it paywalled? Why are you making/have made people across the world sift through the worst material on offer by the wide uncensored Internet to train your LLMs? Why do you have a for-profit LLC operating under a non-profit, or for that matter, a "Public Benefit Corporation" that has to answer to shareholders at all?
Related to that:
> or the needs for hundreds of billions of dollars of compute to train models and serve users.
How does that serve humanity? Redirecting billions of dollars to fancy autocomplete who's power demands strain already struggling electrical grids and offset the gains of green energy worldwide?
> A lot of people around OpenAI in the early days thought AI should only be in the hands of a few trusted people who could “handle it”.
No, we thought your plagiarism machine was a disgusting abuse of the public square, and to be clear, this criticism would've been easily handled by simply requesting people opt-in to have their material used for AI training. But we all know why you didn't do that, don't we Sam.
> It will of course not be all used for good, but we trust humanity and think the good will outweigh the bad by orders of magnitude.
Well so far, we've got vulnerable, lonely people being scammed on Facebook, we've got companies charging subscriptions for people to sext their chatbots, we've got various states using it to target their opposition for military intervention, and the White House may have used it to draft the dumbest basis for a trade war in human history. Oh and fake therapists too.
When's the good kick in?
> We believe this is the best path forward—AGI should enable all of humanity^1 to benefit each other.
^1 who subscribe to our services
Because they're concerned about AI use the same way Google is concerned about your private data.
---
## *What Has Changed*
### 1. *OpenAI’s For-Profit Arm is Becoming a Public Benefit Corporation (PBC)*
* *Before:* OpenAI LP (limited partnership with a “capped-profit” model). * *After:* OpenAI LP becomes a *Public Benefit Corporation* (PBC).
*Implications:*
* A PBC is still a *for-profit* entity, but legally required to balance shareholder value with a declared public mission. * OpenAI’s mission (“AGI that benefits all humanity”) becomes part of the legal charter of the new PBC.
---
### 2. *The Nonprofit Remains in Control and Gains Equity*
* The *original OpenAI nonprofit* will *continue to control* the new PBC and will now also *hold equity* in it. * The nonprofit will use this equity stake to fund “mission-aligned” initiatives in areas like health, education, etc.
*Implications:*
* This strengthens the nonprofit’s influence and potentially its resources. * But the balance between nonprofit oversight and for-profit ambition becomes more delicate as stakes rise.
---
### 3. *Elimination of the “Capped-Profit” Structure*
* The old “capped-return” model (investors could only make \~100x on investments) is being dropped. * Instead, OpenAI will now have a *“normal capital structure”* where everyone holds unrestricted equity.
*Implications:*
* This likely makes OpenAI more attractive to investors. * However, it also increases the *incentive to prioritize commercial growth*, which could conflict with mission-first priorities.
---
## *Potential Negative Implications*
### 1. *Increased Commercial Pressure*
* Moving from a capped-profit model to unrestricted equity introduces *stronger financial incentives*. * This could push the company toward *more aggressive monetization*, potentially compromising safety, openness, or alignment goals.
### 2. *Accountability Trade-offs*
* While the nonprofit “controls” the PBC, actual accountability and oversight may be limited if the nonprofit and PBC leadership overlap (as has been a concern before). * Past board turmoil in late 2023 (Altman's temporary ousting) highlighted how difficult it is to hold leadership accountable under complex structures.
### 3. *Risk of “Mission Drift”*
* Over time, with more funding and commercial scale, *stakeholder interests* (e.g., major investors or partners like Microsoft) might influence product and policy decisions. * Even with the mission enshrined in a PBC charter, *profit-driven pressures could subtly shape choices*—especially around safety disclosures, model releases, or regulatory lobbying.
---
## *What Remains the Same (According to the Letter)*
* OpenAI’s *mission* stays unchanged. * The *nonprofit retains formal control*. * There’s a stated commitment to safety, open access, and democratic use of AI.
Is OpenAI making a profit?
I've been feeling for some time now that we're sort of in the Vietnam War era of the tech industry.
I feel a strong urge to have more "ok, so where do we go from here?" and "what does a tech industry that promotes net good actually look like?" internal discourse in the community of practice, and some sort of ethical social contract for software engineering.
The open source movement has been fabulous and sometimes adjacent to or one aspect of these concerns, but really we need a movement for socially conscious and responsible software.
We need a tech counter-culture. We had one once, but now we need one.
But there are still plenty of mission-focused technology non-profits out there. Many of which have lasted decades. For example: Linux Foundation, Internet Archive, Mozilla, Wikimedia, Free Software Foundation, and Python Software Foundation.
Don't get me wrong, I'm also disappointed in the direction and actions of big tech, but I don't think it's fair to dismiss the non-profit foundations. They aren't worth a trillion dollars, however they are still doing good and important work.
This indicates that they didn't actually want the nonprofit to retain control and they're only doing it because they were forced to by threats of legal action.
Threats of legal action are among the only behavioral signals it can act on while staying in its mandate. Others include regulation and the market.
This is all operating as it was designed, by humans, multiple economic cycles ago.
So were do I vote? How do I became a candidate to be a representative or a delegate of voters? I assume every single human is eligible for both, as OpenAI serves the humanity?
Edit: also apparently known as contronym.
It generally means broadening access to something. Finance loves democratising access to stupid things, for example.
> word is a homonym of its antonym?
Inflammable in common use.
free (foss) -> non-profit -> capped-profit -> public benefits corporation -> (you guessed it)1) You're successful.
2) You mess up checks-and-balances at the beginning.
OpenAI did both.
Personally, I think at some point, the AGs ought to take over and push it back into a non-profit format. OAI undermines the concept of a non-profit.
Checks-and-balances need to be robust enough to survive bad people. Otherwise, they're not checks-and-balances.
One of the tricks is a broad range of diverse stakeholders with enforcement power. For example, if OpenAI does anything non-open, you'd like organizations FSF, CC, and similar to be represented on their board and to be able to enforce those rules in court.
(1) be transparent about exactly which data was collected for the model
(2) release all the source code
If you want to benefit humanity, then put it under a strong copyleft license with no CLA. Simple.
Musk claimed Fraud, but never asked for his money back in the brief. Could it be his intentions were to limit OpenAI to donations thereby sucking the oxygen out of the venture capital space to fund Xai's Grok?
Musk claimed he donated $100mil, later in a CNBC interview, he said $50-mil. TechCrunch suggests it was way less.
Speakingof humanitarian, how about this 600lbs Oxymoron in the room: A Boston University mathematician has now tracked an estimated 10,000 deaths linked to the Musk's destruction of USAID programs, many of which provided basic health services to vulnerable populations. He may have a death count on his reume in the coming year.
Non profits has regulation than publicly traded companies. Each quarterly filings is like a colonoscopy with Sorbonne Oxley rules etc. Non profits just file a tax statement. Did you know the Chirch of Scientology is a non-profit.
He's a symptom of a problem. He's not actually the problem.
But to speak plainly, Musk is a complex figure, frequently problematic, and he often exacts a tool on the people around him. Part of this is attributable to his wealth, part to his particulars. When he goes into "demon mode", to use Walter Isaacson's phrase, you don't want to be in his way.
I'm a citizen, the laws of politics are the problem.
> Musk is a complex figure
Hogwash. He's greedy. There's nothing complex about that.
> and he often exacts a tool on the people around him
Yea it's a one way transfer of wealth from them to him. The _literal_ definition of a "toll."
> When he goes into "demon mode"
When he decides to lie, cheat and steal? Why do you strain so hard to lionize this behavior?
> you don't want to be in his way.
Name a billionaire who's way you would _like_ to be in. Elon Musk literally stops existing tomorrow. A person who's name you don't currently know will become known and take his place.
His place needs to be removed. It's not a function of his "personality" or "particulars." That's just goofy "temporarily embarrassed billionaire" thinking.
> lionize: give a lot of public attention and approval to (someone); treat as a celebrity: modern athletes are lionized.
Where in my comment do I lionize Musk?
Please calm down. Please try to be charitable and curious rather than accusatory towards me.
You attribute to personality what should be attributed to malice. You do this three times.
> Please calm down
I am perfectly calm.
> Please try to be charitable and curious rather than accusatory towards me.
In attempting to explain why my point of view has been misunderstood by you I also attempted to find a reason for it. I do not think my explanation makes you a bad person nor do I think you should be particularly confronted by it.
What have I misunderstood? Help me understand. What is the key point you want to make that you think I misunderstand?
>> (me) When he goes into "demon mode"
> When he decides to lie, cheat and steal? Why do you strain so hard to lionize this behavior?
I hope this is clear: I'm not defending Musk's actions. Above, I'm just using the phrase that Walter Isaacson uses: "demon mode". Have you read the book or watched an interview with Isaacson about it? The phrase is hardly flattering, and I certainly don't use it to lionize Musk. Is there some misunderstanding on this part?
>>>> (me) But to speak plainly, Musk is a complex figure, frequently problematic, and he often exacts a tool on the people around him. Part of this is attributable to his wealth, part to his particulars. When he goes into "demon mode", to use Walter Isaacson's phrase, you don't want to be in his way.
>> (me) Where in my comment do I lionize Musk?
> You attribute to personality what should be attributed to malice. You do this three times.
Please spell this out for me. Where are the three times I do this?
Also, let's step back. Is the core of this disagreement about trying to detect malice in Elon's head? Detecting malice is not easy. Malice may not even be present; many people rationalize actions in such a way so they feel like they are acting justly.
Even if we could detect "malice", wouldn't we want to assess what causes that malice? That's going to be tough to disentangle with him being on the Autism spectrum and also having various mental health struggles.
Along with most philosophers, I think free will (as traditionally understood) is an illusion. From my POV, attempting to blame Musk requires careful explanation. What do we mean? A short lapse of judgment? His willful actions? His intentions? His character? The overall condition of his brain? His upbringing? Which of these is Elon "in control of"? From the materialist POV, none.
From a social and legal POV, we usually draw lines somewhere. We don't want to defenestrate ethics or morality; we still have to find ways to live together. This requires careful thinking about justice: prevention, punishment, reintegration, etc. Overall, the focus shifts to policies that improve societal well-being. It doesn't help to pretend like people could have done otherwise given their situation. We _want_ people to behave better, so we should design systems to encourage that.
I dislike a huge part of what Musk has done, and I think more is likely to surface. Like we said earlier -- and I think we probably agree -- Musk is part of a system. Is he a cause or symptom? It depends on how you frame the problem.
The newer version included sponsored products in its response. I thought that was quite effed up.
o3 still drives really well for me.
If you see a post that ought to have been moderated but hasn't been, the likeliest explanation is that we didn't see it. You can help by flagging it or emailing us at hn@ycombinator.com.
At the same time, though, we need you (<-- I don't mean you personally, but all commenters) to follow HN's rules regardless of what other commenters are doing.
Think of it like speeding tickets [2]. There are always lots of other drivers speeding just as bad (nay, worse) than you were, and yet it's always you who gets pulled over, right? Or at least it always feels that way.
[1] https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
[2] https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
Key Structure Changes:
- Abandoning the "capped profit" model (which limited investor returns) in favor of traditional equity structure - Converting for-profit LLC to Public Benefit Corporation (PBC) - Nonprofit remains in control but also becomes a major shareholder
Reading Between the Lines:
1. Power Play: The "nonprofit control" messaging appears to be damage control following previous governance crises. Heavy emphasis on regulator involvement (CA/DE AGs) suggests this was likely not entirely voluntary.
2. Capital Structure Reality: They need "hundreds of billions to trillions" for compute. The capped-profit structure was clearly limiting their ability to raise capital at scale. This move enables unlimited upside for investors while maintaining the PR benefit of nonprofit oversight.
3. Governance Complexity: The "nonprofit controls PBC but is also major shareholder" structure creates interesting conflicts. Who controls the nonprofit? Who appoints its board? These details are conspicuously absent.
4. Competition Positioning: Multiple references to "democratic AI" vs "authoritarian AI" and "many great AGI companies" signal they're positioning against perceived centralized control (likely aimed at competitors).
Red Flags:
- Vague details about actual control mechanisms - No specifics on nonprofit board composition or appointment process - Heavy reliance on buzzwords ("democratic AI") without concrete governance details - Unclear what specific powers the nonprofit retains besides shareholding
This reads like a classic Silicon Valley power consolidation dressed up in altruistic language - enabling massive capital raising while maintaining insider control through a nonprofit structure whose own governance remains opaque.
I was trying to put all the text into gpt4 to see what it thought, but the select all function is gone.
Some websites do that to protect their text IP, which would be crazy to me if that’s what they did considering how their ai is built. Ha